@context should be required for Schematron rules in ODD

[martindholmes]

Following discussion by the ATOP group and by a larger group of attendees at the TEI Members Meeting in 2023, there is a general consensus that the current practice of allowing Schematron rules to be written without an explicit @context and then having an ODD processor attempt to infer a suitable context for the rule is unworkable. Therefore <sch:rule> should be required when there is <sch:assert> or <sch:report>, and @context should be required on it (as it is by the Schematron schema itself).

[sydb]

@martindholmes and I discussed the caveat that only <constraintSpec scheme="schematron">s that contain an assertion (via either <sch:assert> or <sch:report>) are required to have an explicit context (via <sch:rule> with @context). At first blush, it might seem unnecessary. When would there be a <constraintSpec> that does not contain an assertion, other than one that contains only namespace declarations (which soon enough will be in a separate element altogether — see #2330)? Turns out there are at least two answers to that question:

• Abstract patterns do not have assertions in them. (And soon the TEI Guidelines themselves will be using abstract patterns.)
• On occasion folks use <constraintSpec> to contain only some combination of <sch:let> and <sqf:fix> (which itself contains XSLT). There are 2 cases in the set of ODDs people contributed to ATOP for testing.

[sydb]

Should have been closed when #2513 was merged, no?