Unfortunately, that usually requires spending altogether too much time getting cozy with the data books for the devices in question. Basically, you need to calculate the draw of each device and its duty cycle - which will give you the number of millamp hours for a day's use. After you add a safety factor, and allow for the inefficiencies of the power supply, the solar panel the battery, and the charge controller, you should have a good idea how much power the system must take in each day to continue operating. The solar panel should be sized appropriately.

If you're interested, the newest revision of the bare-bones board has that filter on-board.

This sounds like it would be a better fit for either a shield or a portion of a dedicated daughter/application specific board that also had (for example) a reconstruction filter and an output buffer.

Tx on PC -> Tx on FT232 -> Rx on AVR
Rx on PC -> Rx on FT232 -> TX on AVR

This way the PC sees the output of the microcontroller and the microcontroller sees the output of the PC.

I've worked with a similar circuit -- it's using the internal transistor in the module as an amplifier.

That depends - can you get into the lights out box? If so, tracing out the cable is a matter of using a dvm to follow the wires out to the connector, and taking good notes. If you need help, there's a good reverse engineering tutorial on tinaja.com.