bitreq: Fix tokio-rustls feature gate

[jamillambert]

The code was gated on the dependency instead of the feature alias which
enables other required dependencies. This caused a build failure when
both std and tokio-rustls features were enabled and not the other
required dependencies.

Change the feature gate to the feature alias instead of the dependency.

[tnull]

Thanks, makes sense I think. But if we only now discovered this, it seems we might be lacking test coverage in CI that would have found this? Could we add the missing pieces in this PR?

[jamillambert]

Thanks, makes sense I think. But if we only now discovered this, it seems we might be lacking test coverage in CI that would have found this? Could we add the missing pieces in this PR?

I had a further look into this and it only fails to build with the exact feature set of std and tokio-rustls. There is no practical way to check every combination of features in CI, and I think the current method of checking a sub set is a good compromise. It did still catch the issue, just not in every run.

I also changed the fix to gate on the features async-https-rustls and async-https-rustls-probe, instead of adding async.

[jamillambert]

Audit / Security audit failure is unrelated to this PR and fixed in #565

[tnull]

ACK f4f1c19

[luisschwab]

ACK f4f1c19

[ton-anywhere]

ACK f4f1c19

[tcharding]

FTR I don't know what I"m supposed to do about red CI jobs for cargo audit? I just merged right past them.

[tnull]

FTR I don't know what I"m supposed to do about red CI jobs for cargo audit? I just merged right past them.

Yeah, I think that fine on a per-PR basis. The CI checks are mostly there to ensure that no new known vulnerabilities are introduced. But if it annoys you and it's happening too often (in my experience it's rare) maybe we just reduce it to the cronjob?