tests: mark some additional tests uppercase-hex sensitive

[charles2910]

Since 8.20.0, three more tests started to fail because of the case used to represent hex digits.

Fixes: #440

[jacobmealey]

I wonder if instead of trying to add a keyword every time libcurl changes the case of something, we instead add an ignore_case flag to the affected test cases? I had done to original work for the uppercase-hex work and at the time I thought it was a nice solution, but now I can see it getting a bit annoying.

[vszakats]

I like the ignore-case flag idea, and/or perhaps make the comparison
recognize % encoded bytes and ignore case automatically?

[jacobmealey]

That's even better, should we remove the uppercase-hex as a "feature" as well?

Or should making percent encoded fields lower (or upper case) be part of trurls normalization step?

[vszakats]

That's even better, should we remove the uppercase-hex as a "feature" as well?

Or should making percent encoded fields lower (or upper case) be part of trurls normalization step?

I tend to think that passing it through as-is helps making sure that
consumer apps remain prepared for both casings, improving their
robustness along the way. It also avoid trurl forming an opinion on
casing.

[charles2910]

perhaps make the comparison recognize % encoded bytes and ignore case automatically?

I agree that having a way to ignore the case of percent encoded stuff is the better solution here. Doing that would also allow us to drop the duplicate test cases for uppercase-hex and lowercase-hex. So I'll try to get back to this in the coming weeks and have another MR for it (things are a bit messy right now, so I expect to have free time to really work on it after the second week of june)