refactor(kernel): use real tokenizer for journal budgeting estimates (#58)

Real LLM Usage is already recorded (InferenceCompletedEvent.tokensUsed),
aggregated per-session (MetricsProjection), exposed via correx stats + server
MetricsInspectionService, and rendered in tui-go. The only remaining hardcoded
length/4 estimates were the journal context-entry budget and the compaction
threshold — both pre-injection (no LLM response exists for the journal text), so
estimation is legitimate, but they now use the real tokenizer estimateTokens()
for consistency with sibling context entries instead of length/4.

core/kernel/src/main/kotlin/com/correx/core/kernel/orchestration/DefaultSessionOrchestrator.kt

@@ -699,7 +699,7 @@ class DefaultSessionOrchestrator(
                     compactionService?.let { svc ->
                         val journalState = decisionJournalRepository.getJournal(ctx.sessionId)
                         val journalText = DecisionJournalRenderer().render(journalState)
-                        val tokenEstimate = journalText.length / 4
+                        val tokenEstimate = estimateTokens(journalText)
                         svc.compactIfNeeded(
                             sessionId = ctx.sessionId,
                             state = journalState,

core/kernel/src/main/kotlin/com/correx/core/kernel/orchestration/SessionOrchestrator.kt

@@ -511,7 +511,7 @@ abstract class SessionOrchestrator(
                 content = journalText,
                 sourceType = "decisionJournal",
                 sourceId = "decision-journal",
-                tokenEstimate = journalText.length / 4,
+                tokenEstimate = estimateTokens(journalText),
                 role = EntryRole.SYSTEM,
             ),
         )