Fix BASE table parsing infinite loop memory crash

[RickyDaMa]

When parsing MinMax records, the token declaring the record isn't consumed, and so the parser immediately starts creating another MinMax record, but the token declaring the record isn't consumed, and so the parser immediately starts creating another MinMax record, but the token declaring the record isn't consumed, and so the parser immediately starts creating another MinMax record, but the token declaring the record isn't consumed, and so the parser immediately starts creating another MinMax record, but the token declaring the record isn't consumed, and so the parser immediately starts creating another MinMax record, but the token declaring the record isn't consumed, and so the parser immediately starts creating another MinMax record—

Killed out of memory.

Instead, we now consume the MinMax token so the parser actually progresses. Also speculatively fixes this same pattern in some GDEF table parsing code

[Hoolean]

(We've made an issue for full compilation in the future, too: #1676 😃)

[cmyr]

Can we add a test case in fea-rs/test-data/parse-tests/good? See this README for a quick explanation of how these tests work.

[RickyDaMa]

Test added ✅

[cmyr]

Thanks!