Split 49 modules/suites into independent git repos; untrack from monorepo

Each top-level module/suite folder is now its own private repo on GitHub
(gsinghpal/<name>) and gitea (admin/<name>), with a fresh single initial
commit. The monorepo no longer tracks them (added to .gitignore + git rm
--cached); working-tree files are retained on disk and managed in their
own repos. The monorepo keeps shared root files (CLAUDE.md, docs/, scripts/,
tools/, AGENTS.md, WIP/obsolete dirs) and full history.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

fusion_accounting/fusion_accounting_ocr/services/ocr_providers/__init__.py

@@ -1,3 +0,0 @@
-from . import base
-from . import tesseract_adapter
-from . import manual_adapter

fusion_accounting/fusion_accounting_ocr/services/ocr_providers/base.py

@@ -1,40 +0,0 @@
-"""OCRProvider contract - every backend must conform.
-
-Mirrors the LLMProvider pattern in fusion_accounting_ai. Future adapters
-(Mindee, Google Document AI, Ollama-vision) drop in alongside the default
-tesseract adapter without touching account.move.
-"""
-
-from abc import ABC, abstractmethod
-from dataclasses import dataclass, field
-
-
-@dataclass
-class OCRResult:
-    raw_text: str = ''
-    confidence: float = 0.0  # 0.0–1.0
-    pages: int = 0
-    backend: str = ''
-    error: str = ''
-    metadata: dict = field(default_factory=dict)
-
-
-class OCRProvider(ABC):
-    """Abstract OCR backend. Subclasses implement extract()."""
-
-    name: str = 'base'
-
-    @abstractmethod
-    def extract(self, image_or_pdf_bytes: bytes, *, mimetype: str = 'application/pdf') -> OCRResult:
-        """Extract text from raw bytes.
-
-        ``mimetype`` hints whether to PDF-render (poppler) or image-decode
-        (PIL) the bytes. Implementations should still inspect the byte
-        signature for safety.
-        """
-        ...
-
-    @classmethod
-    def is_available(cls) -> bool:
-        """Return True if the backend's runtime deps are present."""
-        return True

fusion_accounting/fusion_accounting_ocr/services/ocr_providers/manual_adapter.py

@@ -1,13 +0,0 @@
-"""Manual fallback adapter - no real OCR, just marks the document as
-'awaiting manual entry'. Used when no real OCR backend is available
-or when the user explicitly disables OCR.
-"""
-
-from .base import OCRProvider, OCRResult
-
-
-class ManualAdapter(OCRProvider):
-    name = 'manual'
-
-    def extract(self, image_or_pdf_bytes, *, mimetype='application/pdf'):
-        return OCRResult(raw_text='', confidence=0.0, pages=0, backend='manual')

fusion_accounting/fusion_accounting_ocr/services/ocr_providers/tesseract_adapter.py

@@ -1,71 +0,0 @@
-"""Tesseract OCR adapter.
-
-Uses the system tesseract binary via pytesseract, with poppler-backed
-PDF rendering via pdf2image. Inside the container these are pre-installed:
-- tesseract-ocr 5.3.4
-- pytesseract 0.3.13
-- pdf2image 1.17.0
-- poppler-utils
-"""
-
-import io
-import logging
-
-from .base import OCRProvider, OCRResult
-
-_logger = logging.getLogger(__name__)
-
-
-class TesseractAdapter(OCRProvider):
-    name = 'tesseract'
-
-    @classmethod
-    def is_available(cls) -> bool:
-        try:
-            import pytesseract
-            from pdf2image import convert_from_bytes  # noqa: F401
-            from PIL import Image  # noqa: F401
-            pytesseract.get_tesseract_version()
-            return True
-        except Exception as e:
-            _logger.debug("TesseractAdapter not available: %s", e)
-            return False
-
-    def extract(self, image_or_pdf_bytes, *, mimetype='application/pdf'):
-        import pytesseract
-        from pdf2image import convert_from_bytes
-        from PIL import Image
-
-        try:
-            is_pdf = (
-                mimetype == 'application/pdf'
-                or (image_or_pdf_bytes[:4] == b'%PDF')
-            )
-            if is_pdf:
-                pages = convert_from_bytes(image_or_pdf_bytes, dpi=200)
-            else:
-                img = Image.open(io.BytesIO(image_or_pdf_bytes))
-                pages = [img]
-
-            texts = []
-            for p in pages:
-                texts.append(pytesseract.image_to_string(p))
-            full_text = '\n\f\n'.join(texts)
-
-            # Heuristic confidence - tesseract has a per-word conf in
-            # image_to_data, but a length proxy is fine for routing
-            # decisions. Future: use pytesseract.image_to_data for a real
-            # average word-level confidence.
-            conf = min(1.0, len(full_text) / 1000.0)
-            return OCRResult(
-                raw_text=full_text,
-                confidence=conf,
-                pages=len(pages),
-                backend='tesseract',
-            )
-        except Exception as e:
-            _logger.warning("Tesseract OCR failed: %s", e)
-            return OCRResult(
-                raw_text='', confidence=0.0, pages=0,
-                backend='tesseract', error=str(e),
-            )