Back out one of the changes in #26178 to get warnings/warnings.chpl working

[bradcray]

While paratesting my branch tonight, I found that

chplenv/chplconfig/warnings/warnings.chpl

was failing due to PR #26178. Though I'm not that familiar with this logic, it seems that reverting half of the changes in it keeps this test working as expected, though it may suggest we need to do more to both achieve the PR's aim and also keep errors for cases like this.

[bradcray]

@jabraham17: I'll need a post-review merge on this, and for you to check how much of your fix this undoes / whether we can have both the warning and support the case you were striving for.

[DanilaFe]

The warning in question relied (prior to Jade's recent PR) on the fact that splitting a list with more than two elements as a, b = l throws an error. But with maxsplit=1, the list will always have two elements, so the error won't be thrown in that case, and the format warning won't be emitted.

There's a tug between allowing = on the right side of VAR=... and ruling out cases such as A==B, which can be parsed as A = (=B). I think it's reasonable to add a special-case warning for == specifically -- two consecutive equal signs, which would fix this test. In the meantime, reverting the change for the night should avoid nightly testing calamities.

[jabraham17]

I agree with everything Daniel said. This change looks good and I will do a proper fix this morning that allows VAR= ...=... but not VAR==

Thank you both for avoiding a slew of nightly failures this morning :)