test : added edge-case unit tests for crawler _normalize_form

[tmdeveloper007]

Closes #1707.

Summary of What Has Been Done:
Added testing/backend/unit/test_crawler_normalize_form_edge_cases.py with 25 unit tests covering edge cases for _normalize_form in backend/secuscan/crawler.py.

The existing tests in test_crawler_surface_parsing.py cover DELETE method and non-dict input items. This file adds:

Changes Made:
CSRF token detection (5 tests):

• All 5 known CSRF token names (csrf, _csrf, csrfmiddlewaretoken, authenticity_token, __requestverificationtoken) correctly set has_csrf_token=True
• Normal input names do not trigger CSRF detection
• Names containing CSRF substring but not exact matches return False
• Empty inputs list returns False

State-changing via input types (5 tests):

• Password input type triggers state_changing=True on GET form
• File input type triggers state_changing=True on GET form
• Hidden input type triggers state_changing=True on GET form
• POST method without special inputs is state_changing=True
• GET form with text-only inputs is state_changing=False

Password field counting (3 tests):

• Correct count when multiple password inputs present
• Zero count when no password inputs
• Non-dict input items ignored in count

Input count (3 tests):

• Correct total for mixed input types
• Zero for empty inputs list
• Zero and no crash when inputs is not a list

Action URL normalization (4 tests):

• Relative action URL resolved via urljoin
• Absolute URL returned unchanged
• Empty action defaults to page URL
• None/missing action defaults to page URL

Output shape (2 tests):

• Original form fields preserved in output
• page_url always set to passed-in value

Impact it Made:

• Comprehensive edge-case coverage for form normalization security analysis
• Prevents silent incorrect state_changing or CSRF token detection
• Documents urljoin behavior for relative vs absolute action URLs

Note: This task is being handled by tmdeveloper007 — please assign to that account when picking it up.

[utksh1]

Closing: this PR is part of the same mass template-generated low-value test batch pattern. The PRs use repetitive boilerplate bodies, copy-pasted structures, and in several cases manufacture tiny helpers or duplicate existing coverage rather than contributing one focused, meaningful improvement. Please submit a smaller, genuinely curated PR if you want to work on one concrete test gap.