packaging: operator delivery and multi-chain

[mfw78]

Land the operator-facing delivery that ships alongside the repo cut: green CI and CD, multi-chain provider configuration and deployment docs, and container image packaging.

Goal

Give an operator everything needed to run the split system in production. That means continuous integration and deployment that stays green (including the sccache fork-PR fail-open fix), a multi-chain provider map with deployment documentation, and ghcr container image packaging.

Scope

The delivery work happens in step with the physical carve so the freshly split repositories ship a runnable, documented product rather than just source. CI and CD are hardened first, closing the sccache fork-PR failure so pull requests from forks no longer break the pipeline. On top of a green pipeline the operator surface is filled in: a provider map that lets a single deployment target multiple chains, deployment documentation, and packaged ghcr images.

The real Swarm remote-store backend was split out of this epic and deferred: see nullislabs/nexum-runtime#33 (removed from M5). This epic no longer covers it.

Milestone: M5: The gated three-repo split.

[mfw78]

Status after the M5 split. Landed: CI (sccache, #337), the swarm remote-store backend (#151), container-image packaging (the promoted root Dockerfile in PR #485), and green standalone CI on all three repos (L1 nexum-runtime, L2 videre-nexum-module, L3 shepherd). Remaining: operator deployment + multi-chain configuration docs. Leaving open for that doc pass.