spike: graduate the data layer to @gemstack/orm (+ adapters + schema) [#64 Phase 0, not for merge]

[suleimansh]

Phase 0 of #64. A reversible spike, NOT for merge. No publish (all four packages are private: true), no shims, no dependents repointed. The point is to prove the mechanical graduation is clean and to surface what the real move costs, as the concrete artifact for the brand-consolidation decision with the Vike team.

What this does

Copies the agnostic engines out of vike-data and renames them to @gemstack/*, following the @gemstack/ai-sdk playbook (copy -> rename -> [stop here for the spike]):

From (vike-data)	To
@universal-orm/core	@gemstack/orm	engine
@universal-orm/memory	@gemstack/orm-memory	adapter
@universal-orm/drizzle	@gemstack/orm-drizzle	adapter
@vike-data/universal-schema	@gemstack/schema	schema engine

The JS-src-direct build model (ship src/*.js, node --test, no dist) is preserved as-is, not rewritten to the TS->dist model the AI family uses.

Result: green in the gemstack workspace

@gemstack/orm          29 pass
@gemstack/orm-memory   10 pass
@gemstack/orm-drizzle  12 pass
@gemstack/schema      103 pass
total                 154 pass / 0 fail

Findings (the reason to run a spike)

1. The ORM runtime is genuinely framework-agnostic (zero Vike/ORM imports), exactly as Epic: graduate the data layer to @gemstack/orm (+ adapters) #64 claimed. Confirmed, not assumed.
2. Schema is not a "later, lower-priority" graduation. @gemstack/orm's own test suite (and orm-memory's) builds fixtures with defineSchema/mergeSchemas from schema. The runtime is schema-free, but the test suite is not, so schema must graduate with the ORM, not after it. Epic: graduate the data layer to @gemstack/orm (+ adapters) #64 should be updated to reflect this.
3. The Rudder adapter is the one coupled package. @universal-orm/rudder (peer @rudderjs/database) is deliberately excluded; it is the "living binding" decision in Epic: graduate the data layer to @gemstack/orm (+ adapters) #64, not part of the agnostic core.
4. One dangling dev-only edge: core/memory carried a @vike-data/universal-schema dev/types reference with no runtime import. It resolves once schema graduates alongside.

What this spike deliberately does NOT do (Phase 2, after the gate)

• No fresh 0.x / publish
• No re-export shims or living bindings at the old @universal-orm/* names
• No repointing of dependents (vike-data, @rudderjs/orm, playgrounds)
• No @universal-orm/rudder decision

All of that is gated on the brand-consolidation decision with the Vike team (consolidate / parallel brands / defer) per #64. This branch is the evidence for that conversation, and should be closed unmerged once the decision is made and real Phase 2 PRs open.