Doppler Range Map TUI & WebGPU

[x46085]

This was a rather large amount of work in the service of "how fast and compelling of a visual demo can we get with the Aleph's GPU", and a lot of code which resulted in a rather simple answer:

• visually, the demo can be pretty cool, with a high performance TUI and WebGPU stream demo.

• architecturally, we are blocked by only having USB 2.0 access to the PlutoSDR. We need to be able to leverage the Aleph's USB3.0 Super Speed or PCIe lanes in order to transfer the data off the PlutoSDR fast enough to make the GPU actually blow minds.

• timing study:
  Timed pipeline (all times from steady-state measurements):

1. GPIO burst trigger                    ~13 ms
2. sdr.rx() -- blocking IIO read        ~1330 ms  (93% of total)
3. Channel sum (chan1 + chan2)            ~23 ms
4. Reshape into chirps (vectorized)      ~9.5 ms
5. DC suppression (if enabled)           ~2 ms
6. MTI filter (if enabled)               ~8 ms
7. GPU processing (FFT2 + post)          ~58 ms
   - to_gpu (cp.asarray)                   2.5 ms
   - fft2 + fftshift + abs (async)         1.3 ms
   - log10 + clip (async)                  0.4 ms
   - GPU sync                              52 ms
   - to_cpu                                1.5 ms
   - float32 cast                          0.9 ms
8. Total                                 ~1435 ms (~0.7 FPS at 512 chirps)

So as you can see above, the RX/TX from the PlutoSDR takes 93% of the roundtrip time, leaving the GPU largely unbothered and idle waiting for work to do.

Regardless, results from this PR:

Original Phaser Lab Demo of a Doppler Range Map:

mac-range-doppler.mov

Rust TUI Doppler Range Map:

TUI-trimmed.mov

WebGPU Doppler Range Map:

web-trim.mov