fix(derive): Support SubcommandsNegateReqs

[epage]

Before there was no way to make SubcommandsNegateReqs with
clap_derive because it required a required field with a sentinel value
for when the required part was negated. We blocked that.

This turned out simpler than I expected.

This came out of the discussion for #2255 but that issue is more
specifically about the panic, so not closing it.

[kbknapp]

Do we want to go ahead and tackle the full issue in this PR? Cause this fix alone still panics in the case of #2255. Granted, as we discussed in that issue, the developer should use an Option + clap(required = true), but that is not exactly intuitive. And the panic message that is emitted doesn't point them in the right direction either (using the test case in the linked issue):

thread 'required_option_type' panicked at 'app should verify arg is required', clap_derive/tests/options.rs:346:18

We should probably change that to something generic-ish like arg 'X' should be of type Option<T>, and manually add #[clap(required = true)] due to conflicting attributes. This doesn't say what those conflicting attributes are, because I'm not sure we want to try and enumerate all cases or detect which one triggered it, but at least points them in the right direction.

[epage]

Do we want to go ahead and tackle the full issue in this PR

I was wanting to keep the PR focused. Handling the panic involves changes unrelated to the ones I was making.