Feature request: rtk wrapper for az afd (Front Door — high frequency, modest per-call savings)

Real-world numbers from 30 days of Claude Code transcripts (686 sessions scanned, paired Bash tool_use → tool_result by tool_use_id):

Subcommand	Calls	Est. tokens	Avg/call
az afd origin	161	10,915	67
(other az afd)	88	4,108	46
Total az afd	249	15,023	60

Different profile from the other az subcommand groups — the per-call output is already small (60 tokens avg), but the call volume is high (249 calls) because debugging Front Door is iterative (origin show → route show → check WAF → check origin again).

The main bloat per-call is the extendedProperties / kind / systemData / etag metadata that appears on every Front Door resource. Removing those alone saves ~30% on small responses.

Suggested wrapper behavior:

• Drop extendedProperties, kind, systemData, etag, provisioningState: Succeeded from all afd shows
• For az afd origin show/list: collapse privateLinkResource block to a one-line privateLinkType:targetId summary unless -o json requested
• For az afd route show: emit name | hostnames[] | originGroup | rulesets[] summary by default

Lower-priority caveat: the generic ARM JSON pruner (filed separately as #1590) would cover most of the per-call savings here as a fallback. Reasonable to defer this issue and revisit once the pruner is in place — by my estimate the pruner alone would close ~70% of the per-call bloat for az afd.

Methodology note: numbers come from scanning ~/.claude/projects/**/*.jsonl transcripts, pairing Bash tool_use blocks with their tool_result by tool_use_id, stripping any rtk prefix to count like-for-like, and estimating tokens at 4 chars/token.