Warn about signed 8-bit truncation for `add sp, e8` and `ld hl, sp + e8` in RGBLINK

We added this warning in #2055 for RGBASM to address #2050, but for RGBLINK it will require a new patch type, hence a new object file revision number.

I had thought we'd count those as breaking changes, which would postpone this until RGBDS v2.0, but perhaps not.

Start in RGBLINK by tracing how patch types are represented and how object file revision numbers are handled. Compare the existing RGBASM warning from #2055, then determine the compatible revision and warning behavior for `add sp, e8` and `ld hl, sp + e8`; done means RGBLINK warns about signed 8-bit truncation without leaving object files incompatible unexpectedly.

Written by the indexing model from the issue text.