Docs: API coverage roadmap (Reports/PfbApiCoverageRoadmap.md)

[juemerson-at-purestorage]

Adds the API coverage roadmap that the tracking issue #46 and the per-family issues (#30–#45) already reference — so the file lives in the repo and can be reviewed/commented on line-by-line.

What it is: a prioritized development guide for closing FlashBlade REST API coverage gaps, derived from Reports/PfbApiDriftReport.{md,json} (28 REST versions, 2.0–2.27). It groups the raw drift findings into functional cmdlet families, ranks them into priority tiers, and records the rationale.

Highlights:

• Reframes the 282 "parameter gaps": most are the two Fusion-workstream params (context_names, allow_errors), whose exclusion resolves 97 endpoints and leaves ~185 with a real residual gap.
• 12 capability families (A–L) + cross-cutting sweeps, mapped to issues -Limit is ignored under AutoPaginate — paginator has no running-total guard #30–Miscellaneous endpoint coverage (login-banner, smtp PATCH, fs performance, presets/workload, …) #45.
• Records the scope boundary with the Fusion context work (Implement Fusion context_names injection #25 / PR Design spec: Fusion context_names injection #22): only context-execution plumbing is external; Fleet/Realm object management is in scope.

Not for immediate merge — posted so the roadmap is tracked and commentable. Happy to merge once the ordering/priority is agreed (comments on ranking especially welcome, per the discussion with @dmann000).

[juemerson-at-purestorage]

Closing this without merging. The roadmap's content has been folded into #46, which is now
self-contained.

The deciding factor is that this document's top priority is already finished. Sweep S4
("P0-1 — fix half-built writers, treat as bugs") named 17 cmdlets with specific
missing-field counts. Re-measured against a current drift report, all 17 are now zero —
closed by #31 / PR #66. Update-PfbCertificate 25→0, New-PfbNfsExportRule 19→0,
Update-PfbArrayConnection 14→0, Update-PfbTlsPolicy 14→0, and so on. The single
remaining hit anywhere in that table is Update-PfbApiClient's max_role, which is the
deliberately hand-excluded deprecated field tracked as #65 item 4.

Every headline figure has also moved, in both directions:

Metric	This PR	Current
Uncovered endpoints	117	96
Endpoints with parameter gaps	282	438
context_names systemic gap	197	270
allow_errors systemic gap	134	118
Public/ cmdlets	523	542
REST versions analysed	28 (2.0–2.27)	29 (2.0–2.28)

The "not-verified endpoints: 164" metric no longer exists at all — the generator replaced it
with partial-confidence scoring (59 endpoints today).

The general problem is that a hand-maintained snapshot of generated data goes stale silently.
This one did so in ten days, and nothing signalled it. Updating the numbers would just restart
that clock, so the durable fix is to stop hand-maintaining the derivable parts: #46 now carries
only the family→priority ranking and the ownership map, and points at
Reports/PfbApiDriftReport.md for every count.

What is worth preserving from this PR, and is not currently generated anywhere: the
family→priority ranking, and the reasoning for putting correctness bugs in shipping cmdlets
ahead of net-new endpoint coverage. Both now live in #46's body — the S4 rationale was right
even though its data expired.

If a committed roadmap artifact is still wanted, the right form is a generated one: a
coverageRoadmap section emitted by tools/Build-PfbApiDriftReport.ps1 from a small curated
family→endpoint table, so the counts and the "already covered, box closes" transitions
regenerate alongside everything else. Happy to open that as its own issue if you want it.

Reopen if you'd rather keep a committed document and update it instead — no objection, it just
needs the S4 table and all counts struck first.

[juemerson-at-purestorage]

Superseded by the #46 rewrite — see the analysis above.