feat(context,infra): compression pipeline stages 4-5 — token pruning + relevance + ToMe

- build() is now suspend: pipeline runs all stages in fixed order, each gated by level
- TOKEN_PRUNE (level 3): TokenPruner interface + LLMLingua-2 sidecar (sidecars/llmlingua)
  + HttpTokenPruner adapter (fails open if sidecar down); prunes freeform, preserves
  protected spans, skips tier-0 turns when TIER_SPLIT on
- TOME_MERGE (level 8): ToMeMerger collapses near-duplicate freeform turns (Jaccard)
- Stage 5 selection: RelevanceScorer + EmbeddingRelevanceScorer (cosine over Embedder);
  query-conditioned reorder so least-relevant freeform drops first under budget
- [orchestration] compression_level + token_pruner_url config, wired in Main
- suspend ripple fixed across builder callers/stubs
This commit is contained in:
2026-07-01 14:29:56 +04:00
parent e0c222392c
commit 047e2a4070
25 changed files with 509 additions and 65 deletions
@@ -68,7 +68,7 @@ import java.util.UUID
class GraphRerouteTest {
private val passthroughBuilder = object : ContextPackBuilder {
override fun build(
override suspend fun build(
id: ContextPackId,
sessionId: SessionId,
stageId: StageId,
@@ -68,7 +68,7 @@ class NeedsArtifactInjectionTest {
private val capturedEntries = mutableListOf<ContextEntry>()
private val capturingBuilder = object : ContextPackBuilder {
override fun build(
override suspend fun build(
id: ContextPackId,
sessionId: SessionId,
stageId: StageId,
@@ -63,7 +63,7 @@ class SteeringPreemptionTest {
private val capturedEntries = mutableListOf<ContextEntry>()
private val capturingBuilder = object : ContextPackBuilder {
override fun build(
override suspend fun build(
id: ContextPackId,
sessionId: SessionId,
stageId: StageId,