Dedupe opendbc code

[aubsw]

Needed for #2171

Dedupes the code in panda that was cribbed from opendbc.

details:

1. As a convention, we now include headers from opendbc using with the relative path from opendbc.
   E.g. #include "safety.h" now becomes #include "safety/safety.h"
2. Delete duplicated code.
3. Fix MISRA test invocation (now it can see opendbc headers).

why?

1. Conflicts

   The duplicated code was causing issues during compilation when trying to set up the source/header model for [$500 Bounty] Refactor panda to use source + header file structure #2171.

   The duplicated macros defined at opendbc/safety/board/utils.h and board/utils.h were confusing the compiler during preprocessing, e.g.:

panda % arm-none-eabi-gcc -o board/bootstubx-safety-panda.o -c -mcpu=cortex-m4 -mhard-float -DSTM32F4 -DSTM32F413xx -Iboard/stm32f4/inc # [... truncated for brevity]
In file included from /Users/aub/programming/panda-cpy/opendbc/safety/safety.h:6,
                 from /Users/aub/programming/panda-cpy/opendbc/safety/main.c:1:
/Users/aub/programming/panda-cpy/opendbc/safety/board/utils.h:38:21: error: expected identifier or '(' before 'void'
   38 | #define UNUSED(x) ((void)(x))
      |                     ^~~~
/Users/aub/programming/panda-cpy/opendbc/safety/main.c:6:1: note: in expansion of macro 'UNUSED'
    6 | UNUSED(heartbeat_engaged);
      | ^~~~~~

I don't think there's a sensible work-around aside from deleting the duplicated macros outright.

2. Comprehensibility

   The change in import paths for headers clarifies where the code is coming from. May help protect us against namespace collisions in the future.

[aubsw]

@robbederks WDYT of this approach as a first step? This help shrink the diff in subsequent PRs and also makes things a bit easier to reason about, IMO.

I still need to figure out the MISRA checks.

I figured out the issue with the MISRA checks.

[robbederks]

Splitting up the code like this is pretty ugly (especially things like the FAULT list being split over the two repos). The duplicated code we have now in panda / opendbc is obviously bad already, but this seems like a step in the wrong direction to me. Requires a bit more thought / refactoring to abstract away cleanly I think (@sshane any ideas?)

Any way we can finish your refactor without these changes in the meantime?

[aubsw]

@robbederks I will give it a shot. Closing for now.