Like I l said, it's such a narrow use. I've never encountered that need. I don't deny it exists, but then so do many edge cases.
Just seems so obvious to expand it to a specified time limit and beyond QS.
It's a bit like v1 of STRING_SPLIT not having the ordinal, but less useful. Many of us told Microsoft about that before release, but weren't listened to. Eventually, the ordinal was added, but it fails under forced parameterization. Slightly depressing all around.
Query Governor Cost Limit was never useful or extended as you propose, despite things like Parameter Sensitivity being such a common problem.
People want a guarantee on maximum execution time in all sorts of scenarios. No one loves a long running transaction blowing things up (including the version store).
Yes, people can write jobs and other monitoring to fix things up in an ad hoc way, but this is an expensive enterprise product we're taking about here, not Postgres or DuckDB.
By all means include zero time as an option and make the query hint also available to work in QS, but releasing the feature as-is in a major box release is just embarrassing, frankly.
It shows a lack of thought or familiarity with the issues people actually encounter day to day. It's really not good enough.
People put up with the lack of investment in e.g. T-SQL because the core engine is strong. This sort of quarter-baked addition undermines that position.
Unless some large and loud customer is directly funding this work, I can see no reason to release it in such a limited form.