fix(core): parallelize apply_tier_results hooks and align WATCHED_SPANS names

[bug-ops]

Summary

• apply_tier_results (crates/zeph-core/src/agent/tool_execution/tier_loop.rs) processed each tool result's RuntimeLayer::after_tool chain and PostToolUse hook firing sequentially, one index at a time, even though the tier's own tool execution already runs bounded-parallel. A tier with many PostToolUse-matching calls paid N sequential subprocess spawns after already paying for parallel tool execution. The layer/hook phase now runs concurrently across a tier's indices via futures::future::join_all, bounded by the same max_parallel semaphore the tier's tool execution uses.
• MetricsBridge::WATCHED_SPANS (crates/zeph-core/src/metrics_bridge.rs, profiling feature) named three spans (agent.prepare_context, agent.tool_loop, agent.persist_message) that never matched any real tracing span, silently making their on_close handling dead code — only llm.chat was ever observed. Renamed the first two to their real span names (core.context.prepare_context, core.tool.native_loop). agent.persist_message is intentionally left unwatched: its real span fires many times per turn, so bridging it would report the wrong call's duration instead of the first user-message persist that TurnTimings::persist_message_ms is meant to measure — that field stays manually-timed only.

Closes #6128
Closes #6111

Test plan

• cargo +nightly fmt --check
• cargo clippy --profile ci --workspace --all-targets --features "desktop,ide,server,chat,pdf,scheduler,testing" -- -D warnings
• cargo clippy --profile ci --workspace --all-targets --features "desktop,ide,server,chat,pdf,scheduler,testing,profiling" -- -D warnings
• cargo nextest run --config-file .github/nextest.toml --workspace --features "desktop,ide,server,chat,pdf,scheduler" --lib --bins — 13075 passed, 35 skipped, 0 failed
• cargo nextest run -p zeph-core --features profiling -E 'test(metrics_bridge) or test(apply_tier_results)' — 6/6 passed
• New concurrency test (after_tool_hooks_run_concurrently_across_tier_indices) verified non-flaky over 20 runs
• Rustdoc gate (RUSTFLAGS="-D warnings" RUSTDOCFLAGS="--deny rustdoc::broken_intra_doc_links" cargo doc --no-deps --workspace --features "desktop,ide,server,chat,pdf,scheduler")
• Independent adversarial critique of state-mutation safety, error/panic isolation under concurrency, and semaphore bound correctness — verdict: minor, non-blocking
• Independent code review — approved
• CHANGELOG.md updated under [Unreleased]
• .local/testing/playbooks/tier-apply-results-concurrency.md and .local/testing/playbooks/metrics-bridge-watched-spans.md added (main repo)
• .local/testing/coverage-status.md updated in place (main repo)
• Live session verification — not required (no config/user-facing surface changed beyond latency and metric accuracy); scenarios documented in the playbooks above for the next CI cycle

Follow-up (non-blocking, noted by reviewer): a dedicated test for hook/layer failure isolation across concurrent indices would strengthen coverage; current isolation is structurally guaranteed (catch_unwind + fail-open hooks, join_all never short-circuits) and independently verified by two reviewers reading the code, but untested directly.