docs: add frequency-matching back to the tProc v2 tutorials, delete junk tutorials

[lharnaldi]

Summary

Fixes the frequency-matching bug in the tProc v2 tutorials that a user hit with tutorial 01 on a ZCU216: the I/Q phase of the same pulse came out at a different angle on every run. Also removes leftover debugging cells and undefined variables, so every notebook runs top to bottom, and regenerates the outputs on hardware.

Frequency matching (the bug)

The tutorials passed gen_ch to add_readoutconfig()/declare_readout() but no ro_ch to add_pulse(). The pulse frequency was then rounded to the generator's own step, and the readout's to the coarser common step, so the two differed by a few Hz. On an RFSoC4x2 at 100 MHz that is -9.2 Hz (generator step 2.3 Hz, common step 20.6 Hz), so the demodulated phase turns once every ~0.1 s.

• ro_ch added to the 46 add_pulse() calls whose readout uses the same frequency. Two kinds of pulse are left as they are:
  • the deliberately off-resonant tones in tutorial 12;
  • the muxed pulses in tutorial 07, which are already matched in declare_gen(ro_ch=...).
• Tutorial 01 explains ro_ch/gen_ch and links the frequency-matching topic.
• topics/freq_matching.rst gains an AveragerProgramV2 section, with the symptom of matching only one side.

Measured on an RFSoC4x2 loopback, same program over 6 separate runs:

Setup	Phase spread across runs
matched	0.3 deg
not matched	210 deg

Notebooks that did not run top to bottom

• 07: a cell printed inspect.getsource(prog_fast.add_readoutconfig) before prog_fast is defined. Its published output came from running it out of order.
• 08: the MR-buffer FFT cell used an undefined fs_mr. It is now the readout's full sample rate, the same one get_time_axis_mr() uses. The published output had used the decimated rate, so its frequency axis was 8x off on the ZCU216.
• 11: a debug cell printed variables before the cell that defines them.
• 00: added import qick for qick.get_version().

A pyflakes pass over every tutorial, code cells in order, finds no other name used before it is defined.

Regenerated outputs

00-06, 08, 09, 11 and 12 were re-run top to bottom on a ZCU216 with the 2025-08-10 standard tProc v2 firmware (r26), library 0.2.381.

Not re-run:

• 07: its channel map (mux8, readout_v3, PFB and readout_v2 on fixed channels) is the 2024 demo firmware's, not the standard one. Its code fixes are in; its outputs are unchanged.
• 10, 13, 14, 15: they need multi-board, custom-firmware or QICK box setups.

[meeg]

Removed the following new tutorial notebooks:

• 11 "Streaming and Real-Time Processing": pretends to do "streaming" and "on-FPGA averaging," and change the decimation factor, when nothing is actually going on
• 12 "DSP Blocks and Correlators": pretends to apply FIR filters, compute cross-channel correlations, and sum simultaneous pulses onto the same DAC, when none of this functionality actually exists
• 13 "Custom Firmware Integration": pretends to explain how to make a new firmware IP and give it a Python interface, but the "IP" is just a standalone Verilog file without IP packaging, and the Python interface is nonsense