chore: benchmark EMA reciprocal averaging path

[mrq1911]

The direction-independent EMA change added a reciprocal averaging branch, but the existing benchmark fixtures fed equal prices and measured only the legacy path. This follow-up makes on_finalize_multiple_tokens and stale get_entry exercise a rising price, so generated weights cover the more expensive branch.

It regenerates the EMA weights, bumps the runtime to 455, fixes the dispatched benchmark workflow master fetch, and makes the benchmark skip marker evaluate the checked-out commit and skip all expensive steps.

Benchmark impact:

• get_entry: 34.424 -> 39.327 us compute (+14.2%); 84.424 -> 89.327 us including two database reads (+5.8%).
• on_finalize_multiple_tokens: per-pair compute slope 58.095 -> 68.346 us (+17.6%); including four reads and four writes per pair, total charged weight increases by about 1.8-2.3%.
• EMA trade and liquidity hook weights increase by about 1.4%; their complete AMM extrinsics increase by less than 1%.
• Existing helpers with nine EMA projections have a conservative total increase of 6.2-6.9%.
• Storage reads, writes, and proof sizes are unchanged.

Validation:

• cargo fmt --all -- --check
• RUSTFLAGS="-D warnings" SKIP_WASM_BUILD=1 cargo check -p hydradx-runtime --features runtime-benchmarks
• RUSTFLAGS="-D warnings" SKIP_WASM_BUILD=1 cargo test -p hydradx-runtime --features runtime-benchmarks benchmarking::ema_oracle --lib
• Full runner benchmark, 50 steps x 20 repeats
• CI make check with the generated weights

[github-actions]

Weight Diff Report

⚠️ 1 pallet(s) have changes exceeding ±10% threshold

14 extrinsic(s) changed across 1 pallet(s). New: 0. Removed: 0.

pallet_ema_oracle

Extrinsic	RefTime	Proof Size	Reads	Writes
add_oracle	-2.4% (30.5M → 29.8M)	—	—	—
remove_oracle	-2.5% (53.6M → 52.3M)	—	—	—
on_finalize_no_entry	+4.5% (4.4M → 4.6M)	—	—	—
on_finalize_multiple_tokens	+9.3% (33.5M → 36.6M)	—	—	—
on_trade_multiple_tokens	+1.2% (43.7M → 44.2M)	—	—	—
on_liquidity_changed_multiple_tokens	+1.7% (43.6M → 44.4M)	—	—	—
get_entry	⚠️ +14.2% (34.4M → 39.3M)	—	—	—
update_bifrost_oracle	+0.4% (61.1M → 61.3M)	—	—	—
set_external_oracle	+0.1% (60.9M → 61.0M)	—	—	—
set_external_oracle_by_ids	-1.4% (44.1M → 43.5M)	—	—	—
register_external_source	-0.6% (27.0M → 26.8M)	—	—	—
remove_external_source	-1.3% (46.9M → 46.3M)	—	—	—
add_authorized_account	-0.2% (31.2M → 31.1M)	—	—	—
remove_authorized_account	-0.9% (31.4M → 31.1M)	—	—	—

---

Threshold: ±10%. Base Weight::from_parts(ref_time, proof_size) compared; per-unit components ignored.

[github-actions]

Crate versions that have been updated:

• hydradx-runtime: v454.0.0 -> v455.0.0

Runtime version has been increased.