[DOCS] Document `null`/edge case handling for SQL functions

[jrodewig]

We don't consistently document how our SQL functions handle null values and other edge cases (such as negative integers in string function args). We should update the docs for each function so users know what to expect.

Due to the number of functions, we'll likely need to break this up over several PRs. We should prioritize string functions so we can document the changes in #68379

• Aggregate functions [DOCS] SQL: Document null handling for functions #74444
• Grouping functions [DOCS] SQL: Document null handling for functions #74444
• Date/Time and Interval Functions [DOCS] SQL: Document null handling for functions #74444
• Full-Text Search Functions (already done)
• Mathematical Functions [DOCS] SQL: Document null handling for functions #74444
• String Functions [DOCS] SQL: Document null handing for string functions #74201, [DOCS] SQL: Move string fn null/edge handling to arg descriptions #74255
• Type Conversion Functions [DOCS] SQL: Document null handling for functions #74444
• Geo Functions [DOCS] SQL: Document null handling for functions #74444
• Conditional Functions And Expressions (already done)
• System Functions (already done)

CC @astefan

[elasticmachine]

Pinging @elastic/es-ql (Team:QL)

[elasticmachine]

Pinging @elastic/es-docs (Team:Docs)

[matriv]

Similar to #68531. We should also check the way we document other edge case for args (e.g. a negative integer for string functions), to have a consistency across all functions, e.g. should it be marked with NOTE, or should it be with bullet points in the description of the args, or the description of the function.

[jrodewig]

We should also check the way we document other edge case for args (e.g. a negative integer for string functions), to have a consistency across all functions, e.g. should it be marked with NOTE, or should it be with bullet points in the description of the args, or the description of the function.

Thanks @matriv. IMO it's more concise to provide this in the description of the arg. You avoid having to restate the arg name. I can work on making this consistent as part of my changes.

[matriv]

Yep, It's fine to have it there, I added this extra check for those cases, because back then I remember there were a few inconsistencies as well (apart from the null handling).