docs: convert reStructuredText sources to MyST markdown

[timsaucer]

Which issue does this PR close?

Closes #.

Rationale for this change

Phase 2 of the documentation-site refresh started in #1578. With the
modern pydata-sphinx-theme + navigation in place, this PR moves the
content format off .rst and onto MyST .md. The motivation:

• Markdown is the lingua franca of agent-tuned tooling. LLMs trained
  on GitHub and modern docs parse Markdown reliably; reStructuredText
  is a minority dialect that frequently confuses both humans editing
  via PR review and agents reading the source. The Apache
  datafusion-comet sibling project completed the same migration
  recently and reported smoother contributor onboarding.
• MyST is a strict superset of CommonMark with directives for the
  Sphinx features we actually use (toctrees, cross-references,
  code-blocks, admonitions, eval-rst escape hatch).
• The myst-parser extension is already in the docs dependency
  group and was loaded by conf.py even before this PR — switching
  the on-disk format is a low-risk, mechanical change.

This PR stacks on #1578 (theme + navbar refresh). It should land
after #1578.

What changes are included in this PR?

Format conversion (mechanical, via rst-to-myst):

• 33 human-authored .rst files under docs/source/ become 33
  .md files — the user guide, contributor guide, IO subsection,
  common-operations subsection, dataframe subsection, top-level
  index, and links.
• Toctrees, cross-references, code blocks, hyperlinks, admonitions,
  and license headers all round-trip cleanly.

Manual fixes layered on top of the converter output:

• Cross-reference anchors. The converter kebab-cased every
  (label)= anchor (e.g. (io-csv)=), but every {ref} in the
  corpus — including the Python docstrings that sphinx-autoapi
  pulls into the API reference — still uses the underscore form
  ({ref}\CSV <io_csv>`). Rewrite the anchors back to underscore form ((io_csv)=, (window_functions)=, (user_guide_concepts)=, (execution_metrics)=`, etc.) so existing references resolve
  without churning every callsite.
• MyST extensions. Enable colon_fence and deflist in
  myst_enable_extensions (the converter emits these on a few
  files, notably dataframe/execution-metrics.md).
• source_suffix. Keep .rst registered even though no
  human-authored RST remains: sphinx-autoapi generates .rst
  under autoapi/ at build time and Sphinx needs the suffix to
  parse it. The comment in conf.py flags this so a future cleanup
  pass doesn't strip it again.

86 {eval-rst} blocks remain in the converted output. Every one of
them wraps a .. ipython:: directive, which has no first-class MyST
equivalent in our extensions setup. The blocks render identically
and don't block the build. Migrating these to a native MyST exec
syntax is a follow-up that requires either myst-nb or a custom
parser registration — out of scope here.

AGENTS.md is updated so the two .rst paths called out under
"Aggregate and Window Function Documentation" point at the new .md
equivalents.

Are there any user-facing changes?

No behavioral change to the datafusion package — only the source
format of the published documentation. Readers of the rendered site
will not notice the migration; the HTML output is unchanged. Internal
cross-references resolve, the pokemon.csv ipython example on the
landing page and the yellow_tripdata_2021-01.parquet example on
the basics page both still execute.

No api change label — public APIs untouched.

Follow-ups (out of scope for this PR)

• Migrate the 86 {eval-rst} .. ipython:: blocks to a
  MyST-native exec syntax. Requires either pulling in myst-nb or
  configuring a per-language parser.
• Phase 3: multi-version doc publishing (the comet pattern).
• Phase 4: asf-site publishing workflow.