Stop enforcing FIXME issue numbers

[brson]

Our tidy script doesn't allow FIXME to be written without a corresponding issue number. This has just led to us littering the codebase with XXX comments instead. Let's please just stop trying to enforce this policy.

cc #3303

[salemtalha]

Do you think, in the tidy script, we should warn if someone attempts to use XXX from here on out? With or without an issue number that is.

[brson]

@salemtalha Yes, once they are all converted to FIXME, the tidy script should error when it sees XXX, similar to how it treats TODO.

[brson]

I don't imagine this issue will be too controversial, but I'd like to hear other opinions.

[alexcrichton]

Sounds reasonable to me, I don't think it matters so much as long as we canonicalize to one "fix this later" form, and FIXME is fine by me.

[huonw]

Fixed by #11817.

[nikomatsakis]

Hah. Not that I am going to resurrect this policy, but I've actually come to like our old FIXME policy, though I once railed against it. I've found it quite useful to be able to write // FIXME, secure in the knowledge that I will HAVE to address this problem before I am able to commit. Maybe we can add some similar string.

In other news, I now hate // NOTE despite advocating for it originally. It seems to be used exclusively to remind us (incessantly) that there is something that should be adjusted upon snapshot; usually though such cases are marked with #[cfg(stage0)] anyhow, and a comment that we can grep for when we do a snapshot would be sufficient.