[breaking] Fixed regression in library discovery

[cmaglie]

Please check if the PR fulfills these requirements

• The PR has no duplicates (please search among the Pull Requests
  before creating one)
• The PR follows
  our contributing guidelines
• Tests for the changes have been added (for bug fixes / features)
• Docs have been added / updated (for bug fixes / features)
• UPGRADING.md has been updated with a migration guide (for breaking changes)

What kind of change does this PR introduce?
Should fix #1740

What is the current behavior?
Bundled libraries from unused platforms are considered for inclusion

What is the new behavior?
Bundled libraries from unused platforms are ignored

Does this PR introduce a breaking change, and is titled accordingly?
No Yes, see comments below

Other information:
The integration test does not check for the compilation error in #1740, but more simply if the message More than one library found for Wire.h appears in a failed build.

[per1234]

I notice #1565 no longer occurs when using the build from this PR.

I don't know whether there was ever a ruling on whether the behavior described in #1565 was considered desirable.

I'm fine with its discontinuation because the one potential value it had in easing migration to an IDE 2.x which did not provide the traditional bundled libraries is lessened now that the IDE does automatically install the libraries on first run (arduino/arduino-ide#663).

I do recommend documenting this change though because it might cause some user's sketches that have dependencies on libraries from the IDE 1.x installation to no longer compile. It will be easy enough for them to fix it by installing the lost libraries via Library Manager though.

[cmaglie]

I do recommend documenting this change though because it might cause some user's sketches that have dependencies on libraries from the IDE 1.x installation to no longer compile. It will be easy enough for them to fix it by installing the lost libraries via Library Manager though.

as breaking change?

[per1234]

as breaking change?

I think it would be safest to do so.

Even though it was undocumented, perhaps even unintended behavior, it has been the behavior for years now (I just verified that #1565 affects the oldest Arduino CLI version I happen to have on my computer, 0.4.0 from 2019-08-05).

[cmaglie]

Done