Add support for types xs:dateTimeStamp and xs:error

[GuntherRademacher]

This PR adds implementations for the types xs:dateTimeStamp and xs:error.

The implementation of xs:error shares similarities with the earlier proposal to introduce a none type in #2160. However, unlike that proposal, xs:error is implemented as an atomic type, allowing us to reuse most of the existing infrastructure and simplifying the overall implementation.

The original motivation for #2160 was to improve tail-call optimization by disregarding the result type of fn:error, and this benefit is still achievable with the current approach (see the new test case TCOTest.typeCheckOnFnError). However, since the declared return type of fn:error has since been generalized to item()*, it is now necessary to explicitly use cast as xs:error in order to take advantage of this optimization.

[ChristianGruen]

However, since the declared return type of fn:error has since been generalized to item()*, it is now necessary to explicitly use cast as xs:error in order to take advantage of this optimization.

Perhaps we should suggest changing the return type to xs:error. What do you think?

[GuntherRademacher]

Perhaps we should suggest changing the return type to xs:error. What do you think?

In fact I was expecting that, and I think we should. This would also align with what the spec says in 3.2.10. The type xs:error:

The practical uses of xs:error as a sequence type are limited, but they do exist. For instance, an error-handling function that always raises a dynamic error never returns a value, so xs:error is a good choice for the return type of the function.

A number of QT4 test cases would fail when the signature is changed with the implementation in this PR, those would need to be adapted.

101,115d100
< K-ErrorFunc-10
< exactly-one((true(), error()))
< Error : FORG0005: Exactly one item expected.
< Expect: FOER0000
<
< K-SeqExactlyOneFunc-7
< exactly-one((true(), error()))
< Error : FORG0005: Exactly one item expected.
< Expect: FOER0000
<
< K-SeqExactlyOneFunc-8
< exactly-one(( error(), true()))
< Error : FORG0005: Exactly one item expected.
< Expect: FOER0000
<
821,825d805
< K-SeqRemoveFunc-13
< remove(error(), 1)
< Result:
< Expect: any of { FOER0000, XPST0005 }
<
1715,1719d1694
<
< K-FunctionProlog-52
< declare function local:myFunction() as empty-sequence() { fn:error() }; local:myFunction()
< Error : XPTY0004: Empty sequence expected, xs:error found: error().
< Expect: FOER0000