Error err:XD0036 doesn't make any sense for invalid [p:]timeout values

[ndw]

Achim pointed this out in private correspondence.

• The [p:]timeout attribute isn't an AVT, so it's a static error if the value is wrong. (I don't believe there are any use cases that would justify making it an AVT.) It should raise a static error not a dynamic one.
• Even if it was the case that it should raise a dynamic error, err:XD0036 is the wrong one because that's specifically about variable and option values. The [p:]timeout setting is neither a variable nor an option.

Achim and I concluded that err:XS0077 would be more appropriate.

We're just about to publish 3.1. Do we want to try to squeeze this in? I don't believe anyone could reasonably argue that it would require another last call because it doesn't actually produce a change in behavior. But a principled "no, we shouldn't make any more changes to 3.1" isn't unreasonable either.

[gimsieke]

No one will probably object if we squeeze it into 3.1. The long-term benefits outweigh short-term risks (objections by the community, applictions already relying on specific error types/codes, etc.). So go ahead and make the change!

[ndw]

I believe we did this in 3.1