Define C++-compatible pin map for Argon

[rgiese]

Problem

The Particle pinmap_defines.h headers establish pin IDs like SCA with a #define, meaning that any stand-alone use of the token SCA gets replaced with a number by the preprocessor. Other Arduino-based platforms don't do this. Because of Particle's headers, Particle apps cannot directly use some Arduino libraries that use a token like SCA internally, such as:

enum class TicPin
{
  SCL = 0,
  SDA = 1,
  TX  = 2,
  RX  = 3,
  RC  = 4,
};

Solution

We should use C++ to establish constants when we're building in C++. This allows library code to safely use tokens like SCA without preprocessor interference.

Since pinmap_defines.h is also used to compile some nRF platform code in C, we unfortunately need to make the pin mappings available through both C and C++ means, whichever we're compiling with. That said, these defines are unlikely to change, to put it mildly, so I think this poses a negligible maintenance burden.

This PR offers this change for the Argon platform only (since that's what I'm using for my current project). If we believe that this is overall a good idea, I'm happy to make the change for all headers and platforms.

Steps to Test

Build device-os/main with make PLATFORM=argon.

---

Completeness

• User is totes amazing for contributing!
• Contributor has signed CLA (Info here)
• Problem and Solution clearly stated
• Run unit/integration/application tests on device
• Added documentation
• Added to CHANGELOG.md after merging (add links to docs and issues)

[rgiese]

cc: @DavidEGrayson

[avtolstoy]

Thanks for the PR. That's always the problem with defines with 'generic' names. We have a couple of other cases (e.g. LED_XXX) where we should move away from such names that could potentially cause problems for applications.

We will consider applying this change into the next major release (3.x). A few points:

1. The PRs should target develop branch, master always points to the current stable release
2. Maybe we should come up with something for C as well (e.g. use static const instead; enums won't do unfortunately as they are signed and pin_t is unsigned and we cannot change its type easily) and ideally minimize duplication as much as possible when defining pins. While the pinmap does not change, it still leaves a possibility for a mistake. Perhaps this can be overcome with tests.
3. This is a breaking change and may break some applications/libraries which rely on #ifdef PIN_NAME