docs(query-language): align prose tables with BNF and modelStringPattern

[aorzelskiGH]

Summary

Align the prose tables in query-language.adoc with the BNF grammar (grammar.bnf) and the modelStringPattern regex in schema.adoc. Redactional only; no grammar or schema change.

Problem

Several descriptive tables in query-language.adoc contradict the formal language:

• Cast operator table lists entries as $str(...), $num(...), ... — the BNF defines them as str(...) / num(...) / hex(...) / bool(...) / dateTime(...) / time(...) without a $ prefix, and the row for $dateTime is duplicated.
• $aas#submodels field table and example comparisons use submodels without the mandatory [<index>] segment, but the BNF production is submodels ( "[" [0-9]* "]" ) "." <ReferenceClause> and the schema pattern is submodels\[[0-9]*\]\.….
• $aasdesc#submodelDescriptors.* table likewise drops the [<index>] segment in every row.
• The example for the idShort row is wrongly printed as $aas#id (copy-paste from the line above).
• The Limitations bullet for SubmodelElement fields lists only semanticId, idShort, value, valueType but the BNF also allows supplementalSemanticIds and language on $sme#.

Solution

• Remove the duplicate $dateTime row and drop the $ prefix in the cast table.
• Add the [<index>] segment to submodels and submodelDescriptors in the field tables and the matching comparison examples.
• Fix the idShort example.
• Expand the Limitations bullet to cover supplementalSemanticIds and language.

Affected files

• documentation/IDTA-01002-3/modules/ROOT/pages/query-language.adoc

Review notes

• No BNF or schema changes; this is purely a prose consistency fix.
• All fixes are driven by the existing grammar.bnf and modelStringPattern; please diff the tables against the regex in schema.adoc to confirm.

Refs: Review Finding T-11