chore: merge release v1.4.5 to main (#1142)

* chore: bump version to 1.4.5 for hotfix

* fix(#1133): honor model_policy on the claude runtime via alias mapping

model_policy (including the anthropic-fable Claude Fable 5 preset) was
silently ignored on the default `claude` runtime: resolveModelInternal
gated the policy branch on `runtime !== 'claude'`, so agents kept
resolving from model_profile and no warning was emitted.

Wire model_policy into the claude path: map the policy-resolved model ID
back to a Claude Code agent alias (claude-fable-5 -> fable, claude-opus-4-8
-> opus, etc.). A value that is already a bare alias is returned as-is; an
ID with no Claude alias warns once to stderr and falls back to the
configured tier alias rather than emitting an unspawnable ID. Non-claude
runtimes return full IDs verbatim (unchanged) and resolveModelForTier
precedence is unchanged.

Authored against the v1.4.4 base (src/core.cts) for the 1.4.5 hotfix
line — NOT next, where this logic lives in src/model-resolver.cts; the
forward-port to next is a separate follow-up.

Closes #1133

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* chore: promote CHANGELOG for v1.4.5

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "gsd-core",
   "displayName": "GSD Core",
-  "version": "1.4.4",
+  "version": "1.4.5",
   "description": "GSD Core is a meta-prompting, context engineering, and spec-driven development system for AI coding agents.",
   "author": {
     "name": "open-gsd",