Support file/stdin content for interactive card sends

Summary

Sending interactive cards from scripts/agents is currently awkward because im +messages-send accepts --content but not a file-based content flag such as --content-file / --content @file.

For large interactive-card JSON payloads, passing the entire serialized JSON string through a shell argument is brittle.

Environment

• lark-cli version observed: 1.0.30
• OS: WSL Ubuntu
• Brand: Feishu
• Command: im +messages-send --msg-type interactive

Current workaround

CARD=$(python3 - <<'PY'
import json
print(json.dumps(json.load(open('/path/to/card.json')), ensure_ascii=False))
PY
)

lark-cli im +messages-send \
  --as bot \
  --chat-id <chat_id> \
  --msg-type interactive \
  --content "$CARD"

This works, but it is easy to break quoting/escaping for large cards.

Proposal

Support one of these equivalent forms:

lark-cli im +messages-send --msg-type interactive --content-file ./card.json

or:

lark-cli im +messages-send --msg-type interactive --content @./card.json

or stdin:

lark-cli im +messages-send --msg-type interactive --content - < ./card.json

The CLI could read the file/stdin as raw JSON for interactive messages and validate that the payload is valid JSON before sending.

Why this matters

AI agents and CI jobs commonly materialize card JSON as a file first, validate it, then send it. A file/stdin path would be safer and more consistent with other commands that already support file-backed content.

Start at the Go implementation of the im +messages-send command and inspect how its --content argument is handled for interactive messages. Support a file or stdin input path, validate the resulting raw JSON, and confirm that the interactive card is sent without shell-quoting workarounds.