Make _danger-local-https feature additive

[DanGould]

See: https://doc.rust-lang.org/cargo/reference/features.html#feature-unification

"Rust features should be additive. That is, enabling a feature should not disable functionality, and it should usually be safe to enable any combination of features"

Before this change io::fetch_ohttp_keys changed its signature based on the feature flag. I noticed downstream that this creates a feature flag mess in bindings and it's one opportunity to address tech debt.

This begins to address #392

[coveralls]

Pull Request Test Coverage Report for Build 12537462995

Details

• 7 of 16 (43.75%) changed or added relevant lines in 2 files are covered.
• 24 unchanged lines in 5 files lost coverage.
• Overall coverage decreased (-0.4%) to 61.384%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
payjoin/src/io.rs	7	11	63.64%
payjoin-cli/src/app/v2.rs	0	5	0.0%

Files with Coverage Reduction	New Missed Lines	%
payjoin/src/receive/error.rs	1	1.46%
payjoin/src/send/mod.rs	2	92.5%
payjoin/src/receive/mod.rs	5	90.31%
payjoin/src/psbt.rs	7	46.56%
payjoin/src/io.rs	9	36.17%

Totals
Change from base Build 12528252226:	-0.4%
Covered Lines:	2882
Relevant Lines:	4695

---

💛 - Coveralls

[nothingmuch]

the code change looks reasonable, but am missing the correspondence between these changes and the following:

Before this change io::fetch_ohttp_keys changed its signature based
on the feature flag. I noticed downstream that this creates a feature
flag mess in bindings and it's one opportunity to address tech debt.

does this just boil down to splitting things so instead of a cfg gated argument to a fn it's a cfg gated fn? if so, then ACK, but if not then i misunderstood something. either way explicitly describing the problem that existed before this commit would be helpful, as "a flag mess in the bindings" is a bit vague

[DanGould]

does this just boil down to splitting things so instead of a cfg gated argument to a fn it's a cfg gated fn?

It does.

A flag mess in the bindings meant that we needed to have switched function signatures downstream as well. Do you need me to change the commit message?

[nothingmuch]

A flag mess in the bindings meant that we needed to have switched function signatures downstream as well. Do you need me to change the commit message?

oops missed this, thanks for re-requesting review.

i see... i don't think it matters much to change the commit message, it helps me to understand in this case, and in general that's a good thing for commit messages to have, but i also doubt this particular explanation will be very meaningful to anyone reviewing the history in hindsight. i'm ACK, but feel free to do that if for extra cookie points ;-)