docs: make all RST code examples runnable in Python

[wenddymacro]

Several doc pages (index, triple_diff, mmm, business_report, practitioner_*) displayed Python examples referencing undefined data/columns, so copying them into a REPL raised NameError/ValueError. Make each snippet self-contained with the library's data generators, and extend the doc-snippet test to CI-cover these files so the examples stay runnable:

• index.rst: define the hero example's data with generate_did_data
• api/triple_diff.rst: build DDD/panel data with generate_ddd_data / generate_staggered_ddd_data; fix column names (group/partition/time)
• api/mmm.rst: supply the experimental panels for the lift-test and Meridian prior examples
• api/business_report.rst: construct a staggered store panel; drop the deprecated fit(aggregate=) kwarg
• practitioner_decision_tree.rst: build concrete panels for the HAD universal-rollout and SurveyDesign examples
• practitioner_getting_started.rst: mark the narrative summary block as context-dependent (it reads results/pt from earlier steps)
• tests/test_doc_snippets.py: add the seven previously-uncovered RST files to RST_FILES so every doc example is smoke-tested in CI

Methodology references (required if estimator / math changes)

• Method name(s):
• Paper / source link(s):
• Any intentional deviations from the source (and why):

Validation

• Tests added/updated:
• Backtest / simulation / notebook evidence (if applicable):

Security / privacy

• Confirm no secrets/PII in this PR: