fix: update logging levels and improve job URL handling in daily task

[TheTrueAI]

No description provided.

[Copilot]

Pull request overview

This PR reduces noisy HTTP client logging during the daily digest run and refactors daily task job URL handling so URLs map directly to the JobListing objects used for DB upserts and per-subscriber evaluation. It also adjusts tests to better isolate external clients (HTTPX + LLM) and ignores pytest temp directories.

Changes:

• Set httpx/httpcore loggers to WARNING in daily_task.py to reduce log noise.
• Refactor daily task URL→job handling to build url_to_job during search aggregation and generate upsert payloads from it.
• Update tests to mock httpx.Client and create_client(); add pytest-of-* to .gitignore.

Reviewed changes

Copilot reviewed 3 out of 4 changed files in this pull request and generated 1 comment.

File	Description
daily_task.py	Lowers HTTPX logging verbosity; builds url_to_job earlier and uses it to generate DB upsert payloads.
tests/test_bundesagentur.py	Mocks httpx.Client to avoid constructing real clients during provider search tests.
tests/test_app_ui.py	Mocks immermatch.llm.create_client during Streamlit AppTest runs to prevent LLM client creation.
.gitignore	Ignores pytest temp directories (pytest-of-*).

[Copilot]

all_jobs is now only used for the log line + empty check, while the rest of the pipeline (upserts, DB ID lookup, per-subscriber unseen jobs) is driven by url_to_job/location_urls. This can make the "Found %d unique jobs" log misleading (it’s unique by title|company|location, not by URL), and the title/company/location dedupe no longer affects what actually gets processed. Consider either (a) removing all_jobs and basing the count/empty check on url_to_job, or (b) only populating url_to_job/location_urls for jobs that survive the all_jobs dedupe, and updating the log message accordingly.