Async full per-tag browse(): port the type-info tree builder to the async client

[ale-rinaldi]

Async full per-tag browse()

The sync browse() returns the full per-tag symbol tree (name + access-sequence + datatype) by combining the EXPLORE type-info container with the DB list and walking the type tree (#4). The async client (S7CommPlusAsyncClient) still has only the experimental shallow browse() (gijzelaerr#687), which uses _parse_explore_fields (one EXPLORE per DB, name/type per field) — not the type-info tree builder.

Goal

await client.browse() returns the same flat variable list as the sync client — {name, access_sequence, data_type, db_number} — by reusing the transport-agnostic s7/typeinfo.py parser/tree-builder and porting the sync Phase A–E orchestration (_s7commplus_client.browse()) to async.

Notes

• s7/typeinfo.py performs no I/O and is fully reusable as-is; only the orchestration needs an async twin (the per-DB LID = 1 → type-info RID reads, the EXPLORE(ObjectOMSTypeInfoContainer) call, and the recombination into the tree).
• This is a distinct concern from both the experimental shallow async browse and the async transport fix.

Depends on

• Full per-tag browse(): parse the type-info tree (access-sequences) #4 — the sync per-tag browse() / typeinfo tree builder (provides the parser + orchestration to mirror).
• Async client: list_datablocks()/EXPLORE missing IntegrityId tail + fragment reassembly #7 — async transport parity (integrity_tail + fragment reassembly); without it the large, multi-fragment EXPLORE(ObjectOMSTypeInfoContainer) response can't be received.

Validation

Against a live S7-1500: await client.browse() yields the same variable count and the same spot-read results as the sync browse().