Add typemeasure to literalexpr

[dawedawe]

I think that's the PSI structure we talked about.
It's not explicitly using the range of < ... > but it seems to work by advansing the lexer to <

[auduchinok]

@dawedawe Yes, it's just as we've discussed. If FCS doesn't provide actual ranges for < then it probably should. It seems like it has ranges inside its cases, and it should be possible to add Range property. It might be useful for Fantomas as well.

The next step here would be to define grammar rules for the measure kinds:

typeMeasure:
  namedMeasure |
  productMeasure |
  ...

namedMeasure:
  typeUsage<TYPE_USAGE, TypeUsage>;

productMeasure:
  typeMeasure<MEASURE_1, Measure1>
  STAR
  typeMeasure<MEASURE_2, Measure2>;

...

After that you'll need to reflect that in the structure produced by the builder too.

[auduchinok]

@dawedawe Everything looks much better now, thanks! I've also left some comments on the tree structure and tests, please take a look when you have time.

[auduchinok]

@dawedawe This is very nice work, there's just a pair of small issues with the structure/grammar correspondence. I've also left few other comments, but they are not as important.

[auduchinok]

I've just realized: perhaps a type usage wasn't needed here, and having just a type reference would be enough for features like Rename. We should probably try removing it when updating this code after FCS is updated.

[auduchinok]

@dawedawe Great work, thanks!