async CCR-Update enabler (#22 Increment 1)

[next-nf]

#22 Increment 1 — the async Gx CCR-Update enabler. Stacked on #48 (base async-establishment-step3); review/merge the stack in order (#45 → #46 → #48 → this).

The reusable piece that lets an async_m procedure issue a mid-session CCR and await its reply — the mechanism #22 (UE-requested bearer resource modification) and #23 need. No #22 procedure yet; that's Increment 2.

• ccr_update_issue (real Gx) — a pure mirror of the landed ccr_initial_issue: ccr_update_pre_send factored verbatim out of the blocking CCR-Update invoke clause, plus smf_aaa_pcf:ccr_update_issue/4 (the shared issue/5 helper now dispatches per procedure). Returns {Promise, Session_at_send, State_at_send}; the reply routes through the existing {'$reply', Promise} → async_m:handle_reply path; fold_cca is reused unchanged.
• smf_aaa_static async path — the pgw_SUITE AAA mock was synchronous (no Promise, canned post-processed answers), which blocked end-to-end testing of any diameter-async consumer. It now has ccr_update_issue/3/ccr_initial_issue/3 that mint a Promise and deliver the canned answer as a raw ['CCA'|AVPs] {'$reply', Promise} (only Result-Code needs list-wrapping to match handle_cca). A CT proves the full loop: issue → receive → fold_cca → a real {pcc, remove, _} event.

Tests

• smf_aaa_static_async_SUITE (the async-CCR-U loop) + diameter_Gx_SUITE 6/6 (the CCR-U invoke extraction is behavior-preserving) — 7/7 together.
• pgw_SUITE 127/127 (the mock's sync path is untouched); default rebar3 compile green.

Next: Increment 2 — the delete_packet_filters → PTI Delete Bearer procedure, built on this + the CCR-U packet-filter wire encoding.

🤖 Generated with Claude Code

[next-nf]

Superseded by the async-invoke redesign. #49's diameter-bridge CCR-U enabler (ccr_update_issue + the smf_aaa_static async mock path) is replaced by the whole-pipeline async foundation now in #48 — nothing here is carried forward. The #22 filter work lives in #56.