feat(replay): record runtime identity; label compatibility; analyse first divergence

Snapshots, replays and experiments now record the runtime they ran under:
build (module version, VCS commit, modified tree, commit time, Go version,
platform), schema version, a digest of the non-secret configuration that
ignores secret and credential changes, personality version, policy digest,
model routing including per-mind overrides, prompt versions, deterministic
subsystem versions and timezone. A replay compares its identity with the
snapshot's and is an exact-candidate, a different-build (differences
listed) or incompatible (schema differs; the replay stops before running).
The first non-exact step is analysed into events only in the recording,
events only in the replay, and the components they belong to, labelled as
inferred; live model calls are named as their own source of difference.
Migration 0022 is additive; unrecorded identities are reported as such.

internal/minds/config.go

@@ -42,6 +42,11 @@ type CognitiveConfig struct {
 	DreamLLMCalls      int      `json:"dream_llm_calls"`
 	DreamMaxMemories   int      `json:"dream_max_memories"`
 	DreamMaxHypotheses int      `json:"dream_max_hypotheses"`
+
+	// LLMRoles overrides the daemon-wide model routing for this mind:
+	// role → model id. It is what lets two forks of one snapshot run on
+	// different providers while sharing everything else.
+	LLMRoles map[string]string `json:"llm_roles,omitempty"`
 }
 
 // AttentionFeatures is the fixed set of scoring features (spec §6). Weights are