Honda: CANFD long control

[mvl-boston]

Adding long control for Honda CANFD models.

Based on PR2162, rebaseling to current opendbc codebase, with addition of safety code.

• Long tuning is in a separate PR at Honda Bosch - live-learning gas pedal tuning #3014
• This is now testdriving fine with the startup cruise fault workaround at Honda - clear startup Bosch faults #3029

This PR had been in draft for months, due to the cruise faults set at a subsequent drive after long control was enabled.

I suspect the car is detecting the radar tester and throwing a fault at shutdown. I matched every control bit from a factory ACC log upon shutdown, except for the radar tester, and the next drive still faulted. All Honda Bosch models retain prior faults for the first few seconds of a subsequent drive, so I have not found any other way to stop next-drive faulting beyond PR3029.

There are sometimes split-second error messages appearing on the dashboard before clearing, but long control is gated behind alphalong on dev-only, so release drivers will not see this.

Recent drive from a canfd loaner: ad9840558640c31d/00000195--03d04b0d1c

[jyoung8607]

• See if BRAKE_ERROR can be made global/common, it's present in a Nidec DBC as well
• Honda: Safety support for driver regen paddle #2638
• Drive testing
• Why does it fault on quick restarts?

[jyoung8607]

1. Does comma3+red panda usage require bus 4 as written?

That's not necessary, each Panda only knows about what it's hooked up to.

2. This makes startup faults temporary as car does this if engine is restarted too soon after shutdown. Is there another way to buffer flush at shutdown to prevent?

Will have to look and see what's broken.

[jyoung8607]

• Why does it fault on quick restarts?

This could be connected to #2632. Not saying it is for sure, but Panda is dealing with a disruption that can back up queues and affect other buses as well, so I wouldn't spend a bunch more time on it until that's been resolved.

[mvl-boston]

Reopening as this is testing well with the cruise fault workaround linked in the description.

[mvl-boston]

In separate AEB testing, found that radar hud was previously blocked by safety. Changed this PR to allow it through to properly displays that CMBS is off on the instrument cluster. No driving impact noticed in testdrive.