Feature Request: Add AI and Radio fallback modes to prevent mirror crashes during bad weather

[Shlok-Parekh09]

The Problem
Right now, our optical pipeline assumes the Shack-Hartmann Wavefront Sensor (SHWFS) is always giving us a perfect, uninterrupted feed. But in the real world, environmental factors like fog, rain, and sandstorms scatter and absorb the light.
When the weather gets bad, the pipeline still tries to run the standard reconstruction math on those corrupted frames. It ends up spitting out mathematical garbage, which pushes the deformable mirror into an unrecoverable state. Basically, the hardware freaks out, we get a crash, and we lose the connection entirely.

The Solution
We need to add a "quality check" step before the math happens, and introduce two fallback routes depending on how degraded the signal is:

Route A: The AI Fix (For mild fog or scattering)
If the signal is just a bit fuzzy, we immediately pause the standard math so we don't feed bad data to the mirror. Instead, we trigger an AI agent. The AI will look at the degraded frame, predict the missing data based on past frames, and patch the gaps. We then send that cleaned-up data to the mirror, keeping the optical link alive without straining the hardware.

Route B: The Hard Fallback (For sandstorms or total signal drop)
If the signal drops completely or is totally blocked, we need an instant system override. We immediately park the mirror in a safe, neutral state to prevent hardware failure. The system should then automatically switch our connection over to the backup low-frequency radio waves. While on radio, the optical sensor will periodically "ping" the sky in low-power mode, waiting for the weather to clear up so we can switch the laser link back on.