Consider not requiring parentheses for chained comparisons

[huonw]

#558 was merged as a conservative step that restricts a small edge case to gain some correctness assistance, possible features and vastly improved error messages for invalid generic specification via f<a>(b).

We're not sure if the grammar special case is worth this, so we should revisit and possibly relax/revert the rule.

[strega-nil]

I'd rather have chaining operations, like, in #2083, to be quite honest. This alternative is so rarely actually useful (as to be completely useless, imo)

[Centril]

Closing in favor of #2083.