The input must not affect the output, it's a closed loop system, which means that in wide range of input voltage, it will regulate the output to what you configured.
And since i haven't seen such unregulated DC/DC for a long time, i suggest to first understand how do you connect it and what are you measuring. Specifically, you have to measure on the module itself to ensure regulation. The wires must be thick enough to hold charge current. Of course, in the end the measurement should be taken near the phone. If the cable is poor, you may need to raise the DC/DC output voltage.
There is a fair bit of assumption in this as the "datasheet" is not very explicit. You would be much better off if we had a schematic of the actual circuit (you would be able to look up the converter datasheet and as a bonus you would get an insight into designing your own boost converter).
However, I would assume from their graph that they don't rate the U3V12F12 beyond 280mA - so I would not use it for your application. Although there are many, many alternatives (including designing your own board) I am assuming you would rather stick within the Polulu range, and your alternative part seems to do exactly what you want. The graph shows it is still characterised up to 700mA output - so we're happy. But how much current will it pull?
Let's examine the case at 300mA output - which you specified as worst case.
\$Power_{out}=Power_{in}\times\eta\$, where \$\eta\$ is the efficiency of the converter. Looking at the graph, we can see that \$\eta\approx0.88\$ at 300mA out. We also know that \$Power = Voltage \times Current = V \times I\$.
So \$V_{out}\times I_{out} = V_{in} \times I_{in} \times \eta\$ and hence \$12\times 0.3 = 3.7 \times I_{in} \times 0.88\$.
From this, \$I_{in} = \frac{12 \times 0.3}{3.7 \times 0.88}= 1.1A\$
So, for the second converter choice, 300mA (0.3A in my calculations) requires around 1.1A in.
As for LC current spikes - these only occur when reactive loads (such as the L (inductor) and C (capacitor) in the name) cause voltage spikes due to decaying energy fields within them. Your simple Lithium polymer cell will not cause this - all I would do is to put a 2A fuse in series with the cell positive lead, close to the cell in order to protect from overcurrent scenarios.
Best Answer
Yes, you can't convert all the power at the input to the output. There are always some energy loss in the conversion.
By your Pololu module's specification, it has a typical efficiency between 80% to 90%. You can refer to their Typical Efficiency and Output Current section to find the maximum current you can get.
But, this module has input current limit (typically between 1.4 and 2A),
So, with 5V input voltage, the input current can't be more then 1.4A. From the graph, we note the efficiency drop as output current increase. At 500mA, the efficiency drop to nearly 80%, then your input current now is
$$ \frac{12V \times 500mA}{0.8 \times 5V} = 1.5A > 1.4A $$
So, your output current can't be more than 500mA, should be about 450mA or so.