Add image preprocessing pipeline for OCR / template matching

[JE-Chen]

The highest-leverage gap from this round's research: locate_text / ocr_read_structure and match_template fed the raw screen capture to OCR / the matcher — small UI text, dark themes, low contrast and skew wreck both, and there was no preprocessing seam anywhere.

• preprocess_image (AC_preprocess_image): chain named steps grayscale → upscale → binarize → deskew → denoise → contrast (CLAHE), in order. Returns an ndarray; the executor command writes the cleaned image to output_path so it's usable from JSON / MCP / the builder.
• Building blocks: to_grayscale, upscale (scale/interp), binarize (otsu / adaptive_mean / adaptive_gaussian), denoise (non-local means), enhance_contrast (CLAHE), deskew + detect_skew_angle (version-robust minAreaRect normalisation, clamped to ±max_angle).
• Injectable haystack (ndarray/path/PIL) → ndarray, so fully headless-testable on synthetic arrays (incl. a rotated-bar deskew case). Base OpenCV only (CLAHE/fastNlMeans/adaptiveThreshold/warpAffine), lazy imports. Wired through all 5 layers + headless test + EN/Zh docs + WHATS_NEW. Qt-free.

[codacy-production]

Not up to standards ⛔

_{NEW Get contextual insights on your PRs based on Codacy's metrics, along with PR and Jira context, without leaving GitHub. Enable AI reviewer
TIP This summary will be updated as you push new changes.}

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud