Add missing Modula-2 *.def extension

[dkearns]

Modula-2 splits the definition of a module into separate interface and implementation files having extensions *.def and *.mod respectively.

Linguist currently only detects *.mod files. This PR adds support for *.def file detection.

See #3657 and #6451 for related discussions.

I'm not sure why END is in those patterns. Assuming a top down search the MODULE line will obviously be found first but I may be missing something about the implementation.

Description

Checklist:

• [ x] I am adding a new extension to a language.
  • [ x] The new extension is used in hundreds of repositories on GitHub.com
    • Search results for each extension:
      • https://github.com/search?type=code&q=NOT+is%3Afork+path%3A*.def+%2F%5E%5Cs*DEFINITION+MODULE%2F
  • [ x] I have included a real-world usage sample for all extensions added in this PR:
    • Sample source(s):
      • https://github.com/gcc-mirror/gcc/blob/master/gcc/m2/gm2-libs/Environment.def
      • https://github.com/gcc-mirror/gcc/blob/master/gcc/m2/gm2-libs/Environment.mod
    • Sample license(s): GPL
  • [x ] I have included a change to the heuristics to distinguish my language from others using the same extension.

[lildude]

See inline comment

[lildude]

This heuristic will never be used as this is the only user of this extension within linguist which in itself is potentially a problem as .def is an incredibly generic extension used by many things.

If you're 100% certain this regex will only match Modula-2 files, then adding this extension to the generic.yml file will ensure this regex takes effect and is used. If there's a risk this misidentifies other languages, it's best to leave off this extension and use an override when needed.

[dkearns]

Sorry for missing that, I thought I'd tested a failing case but obviously not.

The pattern for *.def is more specific than that applied to *.mod filenames and there are only 30% the number of files on GitHub. I tested as well as I can with GitHub search and didn't find any false positives.

The current matching for *.mod is also very poor. Most of the GNU Modula-2 repository appears to match as AMPL, for example. So, I've also fixed that pattern to limit it to Modula-2 syntax.

Some breaking changes were made to address #3433 but they don't make any sense to me. The issue reporter seems to be using Active Oberon, a different language. They may be asking for these files to match as Modula-2 because the syntax is similar, Oberon being the next in the line of Niklaus Wirth's languages but not recognised by Linguist.

However, Oberon is is no more Modula-2 than it is Pascal. Modula-2 does not support a . in the module name and I can find no examples of any old non-standard Modula-2 dialects supporting that either.

[lildude]

Please add a heuristic test for this extension.

[dkearns]

Please add a heuristic test for this extension.

Done, thanks.

[lildude]

Sorry. I should have been clearer... we need the heuristic test and fixtures as implemented by other generic extensions. See what has been done for the likes of .sol and .cmp.