fix: reject identifier keys containing special characters

[b-per]

Summary

• Adds VALID_IDENTIFIER_RE = re.compile(r"^[a-zA-Z0-9_-]+$") to schemas/job.py as the single source of truth for what characters are valid in a job identifier
• Updates _validate_job_identifiers in the loader to use it, replacing the spaces-only check
• Raises a clear LoadingJobsYAMLError at load time instead of silently failing during the dbt Cloud round-trip

Root cause (Closes #199)

Identifiers are embedded in job names as [[identifier]] and extracted via regex when fetching from dbt Cloud. The extraction regex only allows [a-zA-Z0-9_-] — any other character (e.g. &) causes extraction to return None, making the job invisible to the matching logic. On every subsequent run the tool treats the job as new and creates a duplicate.

The previous validator only rejected spaces; characters like &, @, . passed through silently.

[github-actions]

Coverage Report

File	Stmts	Miss	Cover	Missing
src/dbt_jobs_as_code
main.py	310	160	48%	120–123, 129, 168–169, 200–203, 209, 245–329, 400–405, 418–419, 437–438, 442–444, 461–510, 539–597, 628–673, 678–679, 683
src/dbt_jobs_as_code/client
__init__.py	179	46	74%	16, 53–54, 62, 99–100, 119–120, 139–140, 155–156, 170–171, 187–188, 205, 215, 295–310, 333, 349–350, 360–372, 375–386, 391–400
src/dbt_jobs_as_code/cloud_yaml_mapping
change_set.py	235	30	87%	27–29, 59, 65–66, 155, 212, 224, 247, 251–252, 258–260, 275–278, 334–348, 401–423, 459–473
src/dbt_jobs_as_code/importer
__init__.py	27	7	74%	14–15, 22–27
src/dbt_jobs_as_code/schemas
__init__.py	23	2	91%	68, 78
common_types.py	60	3	95%	67–68, 91
config.py	14	1	93%	18
job.py	132	6	95%	237, 242, 258–267
TOTAL	1183	255	78%

Tests	Skipped	Failures	Errors	Time
167	0 💤	0 ❌	0 🔥	13.170s ⏱️

[b-per]

CI passed