perf(ccusage): reuse usage JSONL scanner for blocks

Blocks no longer need non-usage JSONL timestamps for file ordering, but the block Bun byte path still iterated every line and manually checked whether the current line contained the usage marker.

Reuse the shared usage-line scanner instead. This keeps blocks on the same marker-driven path as daily/session, removes the redundant per-line marker guard, and shrinks the ccusage build from 405.25 kB to 404.44 kB.

Same-host A/B, Apple M3 Pro, real local Claude data, LOG_LEVEL=0, COLUMNS=200, built Bun dist, hyperfine via comma:

- blocks --offline --json: 419.2ms +/- 20.1ms -> 412.0ms +/- 5.8ms (~1.02x)

- blocks --offline table: 420.7ms +/- 21.2ms -> 416.9ms +/- 9.4ms (~1.01x)

Latest JS/Rust #977 reference after this change:

- JS daily/session/blocks JSON: 352.9ms / 368.8ms / 411.4ms

- Rust daily/session/blocks JSON: 245.9ms / 253.1ms / 256.1ms

Validation: pnpm run format; pnpm typecheck; pnpm run test; pnpm --filter ccusage run build; baseline/current JSON parity for daily/session/blocks/monthly/weekly; baseline/current table parity for daily/session/blocks; Bun/Node JSON parity for daily/session/blocks/monthly/weekly.

Author: ryoppippi

apps/ccusage/src/data-loader.ts

@@ -2007,10 +2007,6 @@ async function collectBlockFileResult(
 
 	const processLine = (line: string): void => {
 		try {
-			if (!line.includes(USAGE_LINE_MARKER)) {
-				return;
-			}
-
 			const data = parseUsageDataLine(line);
 			if (data == null) {
 				return;
@@ -2074,28 +2070,7 @@ async function collectBlockFileResult(
 		}
 	};
 
-	const bytes = await readBufferedJSONLBytes(file);
-	if (bytes != null) {
-		const content = Buffer.isBuffer(bytes)
-			? bytes
-			: Buffer.from(bytes.buffer, bytes.byteOffset, bytes.byteLength);
-		let lineStart = 0;
-		let markerIndex = content.indexOf(USAGE_LINE_MARKER_BUFFER);
-		while (lineStart < content.length) {
-			let lineEnd = content.indexOf(10, lineStart);
-			if (lineEnd === -1) {
-				lineEnd = content.length;
-			}
-			const decodeEnd = lineEnd > lineStart && content[lineEnd - 1] === 13 ? lineEnd - 1 : lineEnd;
-			if (markerIndex !== -1 && markerIndex < decodeEnd) {
-				processLine(content.subarray(lineStart, decodeEnd).toString('utf8'));
-				markerIndex = content.indexOf(USAGE_LINE_MARKER_BUFFER, lineEnd + 1);
-			}
-			lineStart = lineEnd + 1;
-		}
-	} else {
-		await processJSONLFileByLine(file, processLine);
-	}
+	await processJSONLUsageFileByLine(file, processLine);
 
 	return {
 		file,