Add a AI/LLM tool policy page

[rouault]

No description provided.

[landryb]

s/gdal/mapserver ?

[rouault]

indeed, thanks

[geographika]

Discussions on mailing list for reference: https://lists.osgeo.org/pipermail/mapserver-dev/2026-June/017368.html

It is still a little unclear to me the following:

LLMs may only be used as an improved auto-completion mechanism, or
for repeated tasks (mechanical refactoring) that could potentially be
completed with a deterministic algorithm.

I think we should allow use for code reviews of PRs (either by the contributor, or leave it open to add a step to the CI in the future), so something like:

LLMs may only be used as an improved auto-completion mechanism, for repeated tasks (mechanical refactoring) that could potentially be completed with a deterministic algorithm, or to assist in identifying potential coding errors. LLM output is advisory only and must not replace human code review or testing.