Edited 2017 - changed recommended long life storage voltage and added comments on fast charging using some recent systems. RM.
What YOU do as regards several of these questions depends largely on what YOU are trying to achieve or test.
Discharge to cutoff is fully discharged (to whatever remaining % that voltage represents). That's the easy one :-)
Percent dropoff of current in tail sets final % of max possible charged reached. There was a superb table given here within last week or so. Can supply later if you don't find it.
Real Men™ plateau at 4.2V and tail down to 10% or even 5% of the constant current rate. This gets the battery full and knocks the stuffing out of it.
Others terminate the current tail at say 25% of cc value.
Optimum lifetime for ongoing usage is at about the end of the constant current phase. That makes it very easy to locate - charge at specified current until desired max voltage is reached, then charge at constant voltage as desired. Here "desired" is to stop immediately. This is the point at which batteries tend to give significantly longer whole of life mAh of storage without grossly reducing mAh capacity per cycle. This is liable to be the point where older "fast chargers" tell you they have finished. Actual % total claimed varies but probably 70% - 80% range.
Newer USB input fast chargers use the term differently. In the case of USB the maximum available charge current at 5V is 5A so that the battery MAY be able to be charged at ~= 6A for the CC part of the cycle using an efficient buck converter to drop voltage and raise current.
[For a buck converter: Vout x Iout = Vin x Iin x efficiency_of_conversion]
Some systems such as QuaqlComms Quick Charge system allow the use of higher charger voltages (9, 12, 20) with specifically designed equipment, so battery charging can be faster for a given voltage provided that the battery specification allows this.
Maximum charge rates for LiIon and LiPo batteries are usually C/1 = 1A per Ah of battery capacity.
At 5V, 5A a USB charger can charge a 6000 mAh 1 cell LiPO battery at max rate - so eg a 10,000 mAh single cell battery used in some larger tablets can not be charged at the allowed 10A ! rate.
For long life storage where actual stored capacity is unimportant, LiIon and LiPo cells should be stored at about 3.7V.
___________________
Using cells without protection adds to the rich tapestry of life. As long as you don't mind the occasional scorch mark on the tapestry that's fine. Note that part of the protection is a one time high capacity fuse under the cap for when things get out of control. Undervoltage discharge destroys. Charging from below a certain voltage at full rate can get fun, I'm told. Charging at reduced rate can bring cell up, I'm told. Below another second level they say don't even think about it. I've had very poor success in trying to get LiIon to misbehave. I have a box of unprotected cells that are very uncooperative about venting with lame etc. Strange. Sony and Apple and even HP seem to be much better at it :-).
LiIon is usually charged at constant current until a max allowed voltage is reached and then is held at that voltage while current tails off under "control" of chemistry of battery until Ichg = k% of Imax where k% is chosen according to longevity or max energy concerns.50% or 25% of Imax gives longer life. 10% or 5% tail gives max capacity but lower life.
Lowering Vpedesatl by 0.1V greatly assists battery life.
Discharging to higher cutoff voltage aids cycle life.
LiIon also has calendar life and starts self destructing from day one so a lightly used battery still dies.
Best cycle life is achieved by stopping charge when Vpedestal is reached and systen changes from CC to Cv. By monitoring voltage this point can be observed. You could even do a "dumb" system that simply watched delta Vbat and declared constant V when delta fell to zero. Only slightly more than a comparator and an RC delay in one input would achieve that.(While Vin is ramping a delayed vin is lower. When Vin pedestals the delayed Vin almost catches up. An offset voltage is needed to allow comparator towork).
LiIon cells mechanically flex the cell as metallic Lithium is "plated" in and out of the cell*. Cycle life is in large part due to battery beating itself to death mechanically.(This is why LiFePO4 lasts much longer and has lower capacity - the material is held in an Olivine matrix that maintains constant shape as active material is moved in/out BUT it takes up some space. )
*Note: Bill Dubuque has suggested that this sentence would be better replaced with " 'LiIon cells mechanically flex the cell as Lithium ions are intercalated".
The distinction is a finer one than may be apparent. However, it is true to say that if you cut open a LiIon cell you would not usually find metallic Lithium in it. Bill notes that this makes primary Lithium cells, which do contain metallic Lithium, a greater fire hazard than LiIon cells.
If you charge a LiIon cell with excessive voltage metallic Lithium will be 'plated out' and "vent with flames" mode usually occurs at about the same time.
Charge to CV level as often and as soon as possible.
If charging all the way their "disconnect message" is a sign of bad ethos. They are probably trying to minimise the risk of fire without telling you.
For longest storage life (as opposed to long life in regular use) storing at a lower voltage than Vmax is in order. Probably at about 3.6 V and only about 30% state of charge. The various Mars Rovers use LiIon batteries and have a design life of about 8000 cycles - but charge to about 3.6 - 3.7 V maximum.
8000 / 365(~=) ~= 22 Terran years.
Best Answer
I could not find that statement in the datasheet you linked. However, take note of this on the input voltage specification:
What this means is that if you want to charge a battery which requires 12.6V, you must provide 13.6V to the unit in order for it to maintain the 12.6V output.
It's a little confusing, because it reads Input (Power Supply Output) but what this section is trying to tell you is that the power management module (PMM) unit's input requirements are dependent on the output voltage desired.
(Put another way, the power supply unit (PSU) must provide 1V more than the battery charging value, so that the PMM can regulate to that voltage.)
This is common for voltage regulators as well. In order to guarantee a stable output voltage, the input voltage must be higher by some amount.