archive: add aggregate extraction byte and entry limits

[dmcilvaney]

archive.Extract limits each regular-file entry to 10 GiB, but it does not limit total extracted bytes or entry count. An archive containing multiple sub-limit files—or a very large number of tiny entries—can still exhaust disk space or inodes during extraction.

This behavior predates archive overlays (introduced in #223), but PRs #275/#276 make Extract part of the source-preparation path.

Expected behavior

• Track cumulative declared regular-file bytes with overflow-safe arithmetic.
• Track total archive entries.
• Fail before either corpus-informed fixed limit is exceeded.
• Preserve the existing per-entry limit.
• Add tests for aggregate bytes and entry count.

This should remain internal policy; no user-facing quota configuration is needed unless real source archives require it.

[Tonisal-byte]

Closed by #275