[Feature]: Cache generated SSTV WAVs, drop output_wav arg, add grayscale modes

[douxxtech]

What changed

• sstv no longer takes an output_wav positional arg (local + server). The WAV path is now derived automatically and cached under <tempdir>/bw_sstv/, based on (image path, mode, image mtime). Re-running the same image+mode reuses the cached file instead of regenerating it (faster).
• shared/sstv.py now pulls in grayscale modes (Robot8BW, Robot24BW) alongside the existing color modes
• Passing an unrecognized mode now logs a warning and falls back to auto-detect instead of failing silently.

Breaking change

Anyone with saved commands/scripts using the old signature:

sstv <image> [mode] [output_wav] [frequency] [loop] [ps] [rt] [pi]

needs to drop the output_wav slot:

sstv <image> [mode] [frequency] [loop] [ps] [rt] [pi]

Old invocations that supplied output_wav will now have that value parsed as frequency and throw ValueError: could not convert string to float.