Crafted item fixes, configurable gems/enchants/Folio, and absolute HPS reporting

[DIzosimov]

Three related pieces of work. Every behavioural default is preserved — the new settings all default to Automatic, which reproduces the previous hardcoded pick exactly.

---

1. Crafted items and embellishments

Embellishments were driven by three separate hardcoded lists that had drifted apart from each other and from EmbellishmentDB, so items were silently dropped rather than scored:

• getItemEffectOptions offered 7 embellishments and excluded Finger/Neck outright, so jewellery embellishments could never be attached. Now derived from EmbellishmentDB via new applicable / slots metadata.
• The SimC importer matched incoming spell names against an allowlist of eight names and discarded everything else — an imported item lost its embellishment and scored as a plain stat stick, with no error. Now resolves against the same DB. The two entry points also disagreed: Sunfire Silk Lining imported but couldn't be added manually; Adorned Fang was the reverse.
• Crafted items carrying an embellishment inherently (Axe-Flingin' Bands, the World Tender's set) have no effect block in ItemDB and simmed as plain gear. Item now attaches those from EmbellishmentDB.setItems, so an ItemDB gap can't silently drop them again.
• Multi-piece embellishment sets are counted once, only when enough pieces are worn, and consume embellishment slots correctly.
• Embellishments with no formula are flagged unmodelled and withheld rather than silently scoring zero. A test fails if a formula lands and the flag isn't removed.

Weapons: buildNewWepCombos only emitted a one-hander when it could pair it with an offhand. With no offhand the weapon was invisible to Top Gear — and if it was the only weapon there were zero combos, zero sets, and an empty report. Unpaired one-handers are now offered alone, with report advice noting the offhand slot is scored as empty.

Embellishment cap: exceeding two embellishments makes every set fail verifySet. Correct, but silent — the item was dropped from all results, or the whole report came back empty. Now reported, warned next to the Go button when it makes a set impossible, and explained in the report.

2. Configurable gems, enchants, consumables and Omnium Folio

Previously the only player choice was the flask.

• Omnium Folio slots 1, 4 and 5 (slots 2 and 3 have a single rune each, so aren't offered).
• Ring and weapon enchants. The weapon choice is a real trade-off — the secondary enchants are budgeted higher than the intellect one.
• Meta gem and stat gem pairing, chosen independently.
• Food, weapon oil and Vantus Rune, which were applied unconditionally with no way to disable them.

Only options with real modelled values are offered. There is no food or enchant data beyond a single hardcoded entry per slot, so those dropdowns are deliberately limited rather than invented.

Bug fixed: the weapon enchant read bonus_stats.mastery while assigning to bonus_stats.haste for Discipline Priest and Restoration Druid, discarding their accumulated haste and replacing it with their mastery.

Bug fixed: settings are grouped by category, but the panel rendered a hardcoded category list — a setting in a new category was stored, wired to the engine, and never drawn, with no error anywhere. The list now lives in one exported place, and SettingsCategories.test.js fails if any category with settings is missing from it, or if any setting lacks a title/tooltip. That check caught a pre-existing empty tooltip on enchantItems.

3. Absolute HPS, Fine Tuning and joint optimisation

hardScore is an intellect-equivalent ranking number, so the report could only show relative differences — and the alternatives list was labelling a hardScore delta as "HPS".

• Sets evaluated through a cast model or ramp sim now carry setHPS, shown on the stat panel with an upgrade percentage over currently-equipped gear, and per alternative alongside the healing and percentage given up.
• The stat-weight path deliberately reports nothing rather than a fabricated number: baseHPS there is a hardcoded placeholder, and that path pre-applies raid buffs the cast models also apply internally.
• The panel no longer shows +0.00% when the best set is simply the player's current gear, which read like the comparison had failed.

Fine Tuning tab prices every gem, enchant, flask and Folio rune against the player's equipped gear by re-running the set with only that setting changed — so the numbers include diminishing returns and the cast model rather than being flat stat-weight estimates. The current pick is identified by what the setting resolved to, since settings normally read Automatic and would otherwise never match an option. Groups where every option prices identically are flagged as unmodelled instead of shown as a real tie (several Folio runes have empty formulas).

Joint optimisation (off by default, since it changes which set wins): the axes interact through diminishing returns, so the best gems depend on the gear and a set ranked second with default gems can win once optimally gemmed. The full cross product is ~27k configurations per set, so this walks coordinate ascent at ~95 evaluations per set instead. Validated against brute force on a real set — identical result, 0.0000% gap. Leading sets are each optimised and re-ranked, then alternatives are re-scored under the winning configuration so they stay comparable.

Also

Two latent crashes where getSetting returns 0 for a missing setting and the result was used as a string (darkmoonHuntStat, flaskChoice — the latter would take out an entire Top Gear run on stale settings). createTopGearWorker moved to its own module so TopGearEngineShared is importable from tests without tripping over import.meta.

Testing

48 suites / 308 tests passing, production build compiles. Adds ~120 tests across six new suites covering the embellishment DB contract, the SimC import path, weapon combinations, the absolute HPS pipeline, settings reachability, and the optimiser (including the brute-force validation).

Verified after merging current dev, not just before.