bump pandaStates to 10Hz

[sshane]

With #29923, fixes #29354 by publishing ignition signal faster to put UI into offroad state.

[sshane]

Full segment size comparison:

• 2Hz pandaStates (49c7b83d348197ed|2023-09-15--16-21-47--1):
  • rlog.bz2: 12,019,885 B
  • qlog.bz2: 879,409 B
• 10Hz pandaStates (49c7b83d348197ed|2023-09-15--16-17-08--1):
  • rlog.bz2: 12,179,348 B (1.32666% increase)
  • qlog.bz2: 911,521 B (3.65154% increase)

[adeebshihadeh]

what's the point of this duplication?

[sshane]

We can also decimate peripheralState in boardd to 2Hz and change what thermald polls for to that service.

[adeebshihadeh]

Why bump perihperalState?

[sshane]

No reason, decimated.

[sshane]

Ran can_replay as fast as possible on the Corolla route and interrupt load was mean 0.292 on master and 0.297 on this branch. Igor says that should be fine.

Another benefit of this change is we don't lose OBD multiplexing flips while fingerprinting, better for debugging errors!

[adeebshihadeh]

Ran can_replay as fast as possible on the Corolla route and interrupt load was mean 0.292 on master and 0.297 on this branch. Igor says that should be fine.

Note that this will be different for USB (C3) and SPI (C3X), but the hit from panda states is way less than can traffic, so doesn't matter anyway.

[adeebshihadeh]

What's the qlog hit now that peripheralState is reverted?

[sshane]

peripheralState had very little impact.

376bf99325883932:
master: 823,951 bytes, 824,038 bytes, 829,203 bytes, 822,070 bytes
10hz: 838,650 bytes, 840,054 bytes, 843,549 bytes, 836,689 bytes (1.81% increase)

49c7b83d348197ed:
master: 910,727 bytes, 910,556 bytes, 907,888 bytes, 913,848 bytes
10hz: 918,186 bytes, 938,407 bytes, 955,519 bytes, 963,478 bytes (3.64% increase)