fix narrowing warning for param definitions

[jpreiss]

When compiling the Python bindings with Clang 11, the param definitions give many noisy warnings of the form:

src/modules/src/controller/attitude_pid_controller.c:328:23: warning: implicit conversion from
      'int' to 'uint8_t' (aka 'unsigned char') changes value from 262 to 6
      [-Wconstant-conversion]
PARAM_ADD(PARAM_FLOAT | PARAM_PERSISTENT, roll_kp, &pidRoll.kp)
~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
src/modules/interface/param.h:118:20: note: expanded from macro 'PARAM_ADD'
    PARAM_ADD_FULL(TYPE, NAME, ADDRESS, 0, 0)
    ~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~
src/modules/interface/param.h:110:35: note: expanded from macro 'PARAM_ADD_FULL'
    { .type = ((TYPE) <= 0xFF) ? (TYPE) : (((TYPE) | PARAM_EXTENDED) & 0xFF), \
    ~                             ^~~~

It appears to be a limitation of Clang: It is worried about storing a value greater than 0xFF in a uint8_t field .type, but it does not recognize that the ternary expression branch will never be hit because of the (TYPE) <= 0xFF check.

It can be fixed by storing TYPE & 0xFF instead. This is a no-op, but since everything's a compile-time constant it shouldn't cause any change in the generated code.

I guess few people are compiling the bindings with Clang, so I understand if you reject the PR because it adds complexity for a niche case.

[knmcguire]

I've assigned @ataffanel for this but he is away at ICRA for the next few weeks so I'll raise it again once he is back

[jpreiss]

Have fun at ICRA!

[ataffanel]

Avoiding warnings is always good! Since this is functionally equivalent I can merge it.

[ataffanel]

Thanks for the PR!