Align Go ingestion boundaries with Python (#16647)

Moves doc_id blob resolution into Parser, tightens chunker/tokenizer to
Python output_format semantics, updates extractor list handling, and
fixes real-template integration tests.

internal/deepdoc/parser/pdf/parse_nocgo.go

@@ -0,0 +1,19 @@
+//go:build !cgo
+
+package pdf
+
+import (
+	"context"
+	"fmt"
+
+	pdf "ragflow/internal/deepdoc/parser/pdf/type"
+)
+
+// Parse is the no-CGO stub for the DeepDOC PDF pipeline. The surrounding
+// parser package converts this into parser.ErrPDFEngineUnavailable.
+func (p *Parser) Parse(ctx context.Context, data []byte, docAnalyzer pdf.DocAnalyzer) (*pdf.ParseResult, error) {
+	_ = ctx
+	_ = data
+	_ = docAnalyzer
+	return nil, fmt.Errorf("deepdoc/pdf: cgo required")
+}

internal/deepdoc/parser/pdf/pdfium/pdfium.go

@@ -1,3 +1,5 @@
+//go:build cgo
+
 // Package pdfium renders PDF pages using libpdfium (statically linked
 // at build time via CGO_LDFLAGS). It exists solely to replace pdf_oxide's
 // RenderPageRaw for use cases where image quality matters for downstream