fix: use json.Marshal instead of json.MarshalIndent for multipart data_json

[dangrondahl]

Summary

• Replace json.MarshalIndent with json.Marshal for multipart form data_json fields
• Pretty-printing inflated payloads ~2x, causing --attestation-data and --user-data to hit the server's 1024 KB per-part limit at ~400-500 KB instead of ~1 MB
• Add test asserting multipart JSON fields contain no indentation

Closes #822

Test plan

• New test TestMultipartFieldJSON_IsCompact verifies compact JSON in multipart fields
• All existing requests tests pass

[claude]

Claude finished @dangrondahl's task in 1m 57s —— View job

---

PR Review: fix: use json.Marshal instead of json.MarshalIndent for multipart data_json

• Phase 1: Static analysis (go build, go vet, go mod tidy)
• Phase 2: Code review (read changed files, assess quality)
• Phase 3: Post review feedback

---

Static Analysis

All three checks pass cleanly — go build ./..., go vet ./..., and go mod tidy produce no issues.

Review Summary

This is a clean, well-scoped fix. The root cause (pretty-printed JSON doubling multipart payload size) is correctly identified, and the one-line fix at requests.go:179 is the right call.

What's good:

• The fix is minimal and targeted — json.Marshal replaces json.MarshalIndent at the single marshalling site
• The follow-up to pretty-print at the logging site only (PayloadOutput) preserves developer experience without affecting wire payloads
• Two focused tests cover both sides: TestMultipartFieldJSON_IsCompact verifies compact wire format, TestMultipartDryRunOutput_IsPrettyPrinted verifies readable debug output
• Clean commit history that follows the repo's TDD convention (fix: then green: commits)

One minor observation (non-blocking, posted as inline comment):

• The value.([]byte) type assertion in PayloadOutput (pre-existing) would panic on unexpected types. A comma-ok guard would be more defensive, though it's safe today given the single call site.

No issues found with correctness, security, or performance. LGTM.