feat(agentkit): add MultiModalRead for images and PDFs

[JonXSnow]

Add MultiModalRead to SandboxTool that returns structured multimodal content for image and PDF files, with fallback to Read for other types.

Images:

• Read raw bytes via readAllBytes Python template (base64 transfer)
• Detect MIME type via magic-number headers (PNG/JPEG/GIF/BMP/WebP/TIFF)
• Enforce 10 MB size limit

PDFs:

• Full read (no 'pages'): return raw PDF bytes, 20 MB limit
• Paged read (with 'pages'): render page range to PNG via go-fitz at 150 DPI, 100 MB limit, max 20 pages per request
• Validate 'pages' parameter syntax and ranges upfront
• Clamp end page with warning when it exceeds total page count

Size enforcement:

• Dual-layer check: Python side (os.path.getsize) rejects before read, Go side re-checks after decode as defense-in-depth
• Use stdout sentinel marker (READALLBYTES_TOO_LARGE) instead of exit codes to distinguish size errors from other failures
• Provide actionable error messages (e.g. suggest 'pages' param when full-read limit is exceeded)

Also:

• Switch Write template to binary mode ('wb') for byte-level fidelity
• Add table-driven tests for parsePagesParam
• Bump go directive to 1.23; upgrade eino to v0.9.0-alpha.5

[github-actions]

Need to create a new tag

The following modules have changes and may need version updates:

• adk/backend/agentkit (Current: adk/backend/agentkit/v0.2.5)

⚠️ Please create and push new version tags for these modules after merging this PR.