The framework no longer populates messages_skipped (removed in microsoft#271),
so the skipped counter and conditional output are dead code.

…or (#268)

## Summary

**Upfront bulk dedup** (commit 1):
- Replaces N per-file `is_source_ingested` DB queries with a single
upfront `are_sources_ingested` bulk query + in-memory `set` lookup
- `_email_generator` now takes a pre-collected `email_entries` list and
`already_ingested: set[str]`, skipping known files before parsing
(avoids wasted parse cost)
- Single `_iter_emails` call feeds both the bulk pre-filter and the
generator
- Removed dead `messages_skipped` tracking from batch callback (always 0
since #271)

**Cleanup**:
- Removes `_flush_skipped`, skip tracking variables, unused
`IStorageProvider` import
- Net -49 lines

## Stack

**This PR is stacked on #267. Merge order: #271 → #267 → #268**

| # | PR | Branch | Description |
|---|---|---|---|
| 1 | #271 | `fix/remove-filter-ingested` | Remove framework dedup |
| 2 | #267 | `feat/vtt-streaming-ingestion` | VTT streaming migration |
| 3 | **#268** (this) | `refactor/email-dedup-consolidation` | Email
bulk dedup |

## Reproducible test

First ingest (populates DB):
```
mkdir -p /tmp/test_eml
# create 2 .eml files (any valid RFC 5322 .eml will do)
uv run python tools/ingest_email.py /tmp/test_eml/ -d /tmp/dedup_test.db -v
```

Re-ingest (should skip all):
```
uv run python tools/ingest_email.py /tmp/test_eml/ -d /tmp/dedup_test.db -v
```

Expected output on re-ingest:
```
Pre-filter: 2 of 2 already ingested

Parsing and importing emails...

Successfully ingested 0 email(s)
Ingested 0 chunk(s)
Skipped 2 already-ingested email(s)
Extracted 0 semantic references
Total time: 0.0s
```

Key behavior: the bulk query fires once before the generator runs. No
emails are parsed on re-ingest (0.0s total time).

## Test plan

- [x] `make format check test` passes (696 tests, 0 pyright errors)
- [x] Manual test: re-ingest same .eml files, verify all skipped via
upfront bulk dedup
- [x] Manual test: ingest with `--verbose`, verify skip reporting in
summary

---------

Co-authored-by: Guido van Rossum <gvanrossum@gmail.com>