fix(sink): batch transformed raw data

[ngjaying]

Summary

• Support batching sink data produced as pre-encoded RawTuple values by dataTemplate.
• Preserve format-specific batch framing for JSON, delimited, and URL-encoded output.
• Document how sendSingle, batching, dataTemplate, and format interact.

Why

• A sink using both batchSize and dataTemplate could fail at runtime with unknown data type: *xsql.RawTuple.
• The streaming batch pipeline transforms each input before the batch writer flushes it, but the batch writer previously accepted only structured rows and tuple lists.
• Planning raw versus structured output up front keeps the runtime path deterministic and lets each format writer frame pre-encoded template results without encoding them again.

Changes In This PR

• Add the optional RawConvertWriter capability and implement it for JSON, delimited, URL-encoded, and fallback stack writers.
• Let TransformOp report whether it produces raw output and select raw/common batch writer mode during sink planning, including the slice-tuple exception.
• Fall back to StackWriter during planning when a selected writer does not support raw output.
• Transfer flushed buffer ownership so later writer reuse cannot mutate emitted batches.
• Add deterministic ownership regression tests across writer reuse cycles.
• Add unit, race, and FVT coverage for raw batching, planner output mode selection, legacy writer fallback, and REST sink output.
• Add English and Chinese documentation with processing-order and format-framing guidance.

Notes

• Existing ConvertWriter implementations remain source compatible because RawConvertWriter is optional.
• In batch mode, dataTemplate describes one batch element rather than the completed batch. Template authors remain responsible for producing output compatible with the configured format.
• JSON adds commas and an outer array, delimited output adds newlines, and URL-encoded output adds & between transformed elements.
• Validation performed:
  • go test ./internal/topo/node ./internal/topo/planner ./internal/converter ./internal/converter/json ./internal/converter/delimited ./internal/converter/urlencoded ./pkg/message ./pkg/modules
  • go test -count=1 -race ./internal/topo/node ./internal/topo/planner ./internal/converter ./internal/converter/json ./internal/converter/delimited ./internal/converter/urlencoded
  • go test ./fvt -run '^TestRuleSuite$/^TestBatchDataTemplateJSON$' -count=1 -v
  • Markdown lint for the modified English and Chinese sink documentation

[codecov]

Codecov Report

❌ Patch coverage is 79.27928% with 23 lines in your changes missing coverage. Please review.
✅ Project coverage is 70.84%. Comparing base (e0ad968) to head (d0fc448).

Files with missing lines	Patch %	Lines
internal/topo/node/batch_writer_op.go	54.84%	10 Missing and 4 partials ⚠️
internal/converter/urlencoded/writer.go	83.78%	4 Missing and 2 partials ⚠️
internal/converter/delimited/writer.go	82.35%	2 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4093      +/-   ##
==========================================
+ Coverage   70.79%   70.84%   +0.05%     
==========================================
  Files         463      464       +1     
  Lines       54339    54443     +104     
==========================================
+ Hits        38466    38568     +102     
- Misses      12887    12888       +1     
- Partials     2986     2987       +1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.