Enable raw acquisition for rfsoc

[rodolfocarobene]

As title says, this PR should implement raw acquisition (for pulse sequences) for the rfsocs.
It should be ok, but I'm waiting to test it.

Checklist:

• Reviewers confirm new code works as expected.
• Tests are passing.
• Coverage does not decrease.
• Documentation is updated.

[codecov]

Codecov Report

Patch coverage: 42.85% and project coverage change: -0.08 ⚠️

Comparison is base (088733d) 59.32% compared to head (0da1807) 59.24%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #480      +/-   ##
==========================================
- Coverage   59.32%   59.24%   -0.08%     
==========================================
  Files          37       37              
  Lines        6203     6211       +8     
==========================================
  Hits         3680     3680              
- Misses       2523     2531       +8     

Flag	Coverage Δ
unittests	59.24% <42.85%> (-0.08%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/qibolab/instruments/rfsoc.py	55.21% <42.85%> (-1.76%)	⬇️

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[rodolfocarobene]

This PR is ready, but there are a few things to note:

• coverage does decrease, but I'm addressing the problem of low coverage (that at the beginning was caused by the server) in a different PR (Time sweepers rfsoc #474) so I would fix this there.
• I tested the routine on hardware using a loopback configuration, connecting directly readout and feedback. Eventually the new feature can be tested using Add optimal integration weights routine qibocal#383 (in particular the tof routine)
• At the moment I didn't remove the changes like Dict->dict
• This should be merged after the qibosoq one

[alecandido]

Apart from the discussion on new-style type hints, this PR seems good to go.

[Jacfomg]

Have you tested if it only allows a single waveform to be acquired ? Maybe it is useful if 2 fit in memory for readout crosstalk analysis ?

[rodolfocarobene]

Thanks @Jacfomg, you are right and I'll try!
I could probably add something more sofisticated than just counting the readout pulses... In fact also single readout pulse, very long, could fill the memory...
In any case, I think all of these are just general improvements and not really required in the short term

[stavros11]

Thanks for implementing this. If you want to merge this now, you can transfer the discussion about multiple waveform acquisition to an issue (if you are planning to look at it later).

FYI, with QM could also not acquire shots of raw waveforms (AveragingMode.SINGLESHOT doesn't work with AcquisitionType.RAW), so for nshots > 1 I could only get the averaged waveform. However, I don't remember for sure, but I believe I could get multiple waveforms by playing multiple readout pulses (on different qubits) at the same time.