This is quite the multipart and deep question. You seem to understand the basics but just in case, I’ll recommend this website as a good overview, albeit dated in terms of current ICs and BMS competitors.
http://liionbms.com/php/index.php
Chips
This is dependent on application. For small packs like the one you’ve drawn, there are a wide variety of chips available. For larger packs, Linear (LTC6803) and Maxim (MAX11081) are the two main competing suppliers of BMS solutions. They provide the most integrated solutions for multi-cell management. The main differences from smaller packs are the higher series voltages, daisy chain communication and noise immune signaling.
Techniques
In either case, voltage monitoring, temperature monitoring and active balancing tend to be the primary needs of a BMS. Other features such as redistribution tend to be less critical and often do not warrant the added cost and hardware.
Solutions
Unfortunately, even though nearly all BMS designs have the same goals, rarely is there an all in one solution. Solutions are based on number of cells, voltages (chemistry), tolerance for loss, communication method etc. These variables are not all standardized between one project and another. Furthermore, technologies keep changing. Linear is considering replacing CAN-BUS with shielded Ethernet. ADC accuracy is increasing. Sample rates are increasing.
The tried and true method is just simple active balancing. That is extra voltage on a cell is bled off resistively. Other than that, the rest of the system is a simple matter of reading all the cell voltages efficiently.
The Future
Eventually, we will see a more robust landscape with better standardization. Companies like Elithion or Nuvation are designing BMS solutions that are nearly drop in solutions. Chip designers are getting a better understanding of customers needs and have already sought to distinguish between the different types of li-ion needs based on different industries. This will mean that we’ll start only paying for features we need. Even battery cell manufacturers are standardizing cell types. For now though, any battery design remains quite customized.
Most buck converters (including this one) have an internal voltage reference and a feedback loop, which regulates the output voltage. The output voltage is set by the resistors R1 and R2. The output voltage is independent of the input voltage fluctuations. The output voltage formula is not related the optional filter.
Switch mode converters can be layout sensitive, so it's usually a good idea to follow a reference design. On the other hand, this one has a top frequency of only 250kHz, so it may be more forgiving than switchers with higher frequencies.
There's more details about the principles of operation of buck converter here and here.
edit: A somewhat odd thing, however, is that the resistor values in the drawing don't quite check with the output voltage
\$V_{out}=1.25\left( 1 + \cfrac{R_2}{R_1} \right)=1.25\left( 1 + \cfrac{3.8}{1.2} \right) = 5.2 V\$
should be 5.0V. I wonder if there's a reason for the extra 0.2V?
Best Answer
It's just a bog-standard series-pass linear reg with with a FET instead of a BJT.
As @PlasmaHH notes, it is often used a pre-regulator in more sophisticate setups. It allows you drop some voltage (and thus power/heat) on an external element. Doing this also lets you extend the voltage input range of your more expensive IC regulator. TI has a separate appnote in which they suggest this.
This prereg idea (even with BJTs) not at all an uncommon. Cordless phones do this a lot for instance because they also use the [higher] pre-regulated voltage directly for a few components and they also derive a lower stable voltage for most of the digital parts. So in those setups it serves a double function (more bang for the buck).