fix(serialization-form): decode collection elements exactly once

[RKS (rksharma-owg)]

Fixes #788.

Form collection readers corrupt literal plus signs, percent sequences and encoded commas. Writer-generated repeated fields can lose values and element boundaries.

The constructor decodes the node before collection readers split it. Primitive conversion then decodes again; enum child nodes also decode the already-decoded values. Split the retained raw collection value first, then decode each primitive or enum element once. Raw comma-separated lists, repeated-field order and duplicates remain supported.

Added direct encoded-value and real writer/factory round-trip regressions for primitive and enum collections, with plus/percent/comma, Unicode, space, repeated-value and raw CSV controls.

Validation: five new cases fail before the correction; all 63 form tests pass locally after it. YAPF, isort, Pylint and mypy pass. Pylint prints the existing suggestion-mode configuration diagnostic (exit 0), also reproduced on unchanged main and upstream CI.

Fork validation: run 38161550435, tested commit 22893db6a351c05335d9913ce3f61d633e4d5ae7, target 79370e7ee8171a42abc828376dcbc331c27623e8, fork-only workflow revision c517af083bb74af2a9877dc97eef434f8e014d70.

All 46 jobs for this commit passed: eight packages across Python 3.10–3.14 with upstream formatting/import/Pylint/mypy/pytest commands, generated-code type checks on all five versions, and CodeQL. Each form run passed all 63 tests with no skips. CodeQL reported zero findings. Upstream import normalization is reproduced; the changed form package additionally needs no normalization. The validation workflow is separate from this PR.

The unchanged target passed upstream build run 37963997486; its form suite ran 55 tests successfully on each supported Python version.

[Copilot]

🟢 Approval recommended

The focused implementation preserves collection boundaries and is well covered by regression tests.

0 open findings

What changed in this PR

Fixes form collection parsing so raw values are split before each element is decoded exactly once.

Changes:

• Corrects primitive and enum collection decoding.
• Adds encoded-value and writer/factory round-trip regressions.

File	Description
form_parse_node.py	Splits raw collection values before decoding.
test_form_parse_node.py	Covers encoded and repeated collection values.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Vincent Biret (baywet)]

Thanks for the contribution!

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[Copilot]

🟢 Approval recommended

The focused fix preserves raw CSV behavior and is covered by relevant regressions.

0 open findings

🧠 Review effort: Balanced