fix: normalize OpenAI token counting, add branch source tracking

pi-ai:
- Fixed usage.input to exclude cached tokens for OpenAI providers
- Previously input included cached tokens, causing double-counting
- Now input + output + cacheRead + cacheWrite correctly gives total context

coding-agent:
- Session header now includes branchedFrom field for branched sessions
- Updated compaction.md with refined implementation plan
- Updated session.md with branchedFrom documentation

packages/coding-agent/src/session-manager.ts

@@ -20,6 +20,7 @@ export interface SessionHeader {
 	provider: string;
 	modelId: string;
 	thinkingLevel: string;
+	branchedFrom?: string; // Path to the session file this was branched from
 }
 
 export interface SessionMessageEntry {
@@ -430,6 +431,7 @@ export class SessionManager {
 			provider: state.model.provider,
 			modelId: state.model.id,
 			thinkingLevel: state.thinkingLevel,
+			branchedFrom: this.sessionFile,
 		};
 		appendFileSync(newSessionFile, JSON.stringify(entry) + "\n");