Rebuild: a lean overview scan for the edge worker

The worker reads the XOVER reply's text directly: fields by position, the
owner from the Xref without arrays (crossposts leave before any subject
parsing), a subject parsed once per file (cached without its segment
counter), a file's details as one shared object, dates as numbers.
Garbage collection had been ~22% of the time.

Same results as parseOverviews + segmentFromHeader on 800k real lines.
edge:bench (one core): boneless 137k → 193k headers/s (CPU 21 → 12.8 s),
ath (2013) 143k → 303k; peak RSS 1.15 / 0.87 GB.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commands/edge_node.ts

@@ -2,7 +2,6 @@ import { BaseCommand, flags } from '@adonisjs/core/ace'
 import type { CommandOptions } from '@adonisjs/core/types/ace'
 import { setTimeout as sleep } from 'node:timers/promises'
 import { gzipSync, unzipSync } from 'node:zlib'
-import { parseOverviews } from '#lib/nntp/index'
 import { RangeIndexer, decodeFrames, toWire, type EdgeBatch, type Lease } from '#lib/edge/index'
 
 const IDLE_MS = 30_000
@@ -81,7 +80,7 @@ export default class EdgeNode extends BaseCommand {
     }
     try {
       for await (const frame of decodeFrames(await api.stream(range.id))) {
-        indexer.add(parseOverviews(unzipSync(frame.data).toString('utf8')), frame.to + 1)
+        indexer.addText(unzipSync(frame.data).toString('utf8'), frame.to + 1)
         chunks++
         const last = frame.to >= range.end
         if (last) indexer.finish()