[Revert] Revert "[CI] Add .dockerignore and increase pip --timeout to 1200s" (#14964) (#15236)

Revert of PR #14964 (merged onto `main`).

Original PR: #14964
Original author: @JavaPythonAIForBAT
Merge commit: `76421f173d68d9f2dae02b43b93cfee0331fa2bf`

---
## What this PR does / why we need it?

Two changes to fix CI image-build failures and speed up builds:

### 1. Increase pip --timeout to 1200s

**Problem:** CI builds failed with `ReadTimeoutError` when pip
downloaded `torch-2.11.0+cpu` wheel (190MB) from `cache-service` → R2
CDN (`download-r2.pytorch.org`).

**Data:**
- R2 CDN download speed from cluster: **~158-233 KB/s**
- torch wheel size: **190MB** (x86_64) / **148MB** (aarch64)
- At 158KB/s, download takes ~20min; pip default timeout is **15s** →
guaranteed failure

```
ERROR: ReadTimeoutError("HTTPConnectionPool(host='cache-service.nginx-pypi-cache.svc.cluster.local', port=80): Read timed out. (read timeout=15)")
```

**Fix:** `--timeout 1200` (20min) on all pip install commands across all
8 Dockerfiles.

### 2. Add .dockerignore

**Problem:** Build context upload to buildkitd took ~26 minutes.

**Data:**
- Total repo: **134MB** (incl. `.git` 92MB)
- After .dockerignore: **50MB** (-63%)
- Upload time: ~26min → estimated ~10min

**Safety verification (all build-critical files retained):**

| Critical path | Files | Retained |
|---------------|-------|----------|
| `vllm_ascend/` | 505 | ✅ |
| `csrc/` (custom kernels) | 1438 | ✅ |
| `setup.py` / `pyproject.toml` / `CMakeLists.txt` | 3 | ✅ |
| `cmake/` + third_party patch files | 3 | ✅ |
| `requirements.txt` / `README.md` | 2 | ✅ |

Excluded files are only: `.github/`, `docs/`, `examples/`,
`benchmarks/`, `tests/`, `.agents/`, `.claude/`, `.gemini/`, `.git*`,
and markdown/docs — none needed at build or runtime.

## Does this PR introduce _any_ user-facing change?
No

## How was this patch tested?
- Verified dockerignore: 3373 git-tracked files → 1349 excluded (all
safe), 2024 retained (all critical)
- CI: torch wheel pre-cached in cache-service + `--timeout 1200`
resolves download timeout

- vLLM main:
https://github.com/vllm-project/vllm/commit/ba07e4a48fc951300d97eb506217dd530583dea3

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>