chore(logging): demote noisy per-region/per-tenant info logs to debug

[jychp]

Type of change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Refactoring (no functional changes)
• Documentation update
• Other (please describe):

Summary

Reduce log volume in cartography by reserving INFO for sync-level start/complete and per-account recap messages, and demoting per-region / per-tenant step logs (fetch, get, transform, retrieved, ...) to DEBUG. The AGENTS.md and docs/root/agents/create-module.md examples already document this convention; this PR brings the rest of the codebase in line.

Highlights:

• cartography/graph/job.py — "Finished job" → DEBUG (fired ~620k/day across cleanup/relationship sub-jobs).
• cartography/intel/aws/dynamodb.py — removed the 8 "Loading DynamoDB ... into graph" calls; they duplicate the aggregate message already emitted by cartography/client/core/tx.py.
• cartography/intel/aws/cloudtrail_management_events.py — per-region Processing/Fetching/Retrieved/Transforming/Loaded calls → DEBUG. The per-account aggregates ("Syncing N regions ...", "... sync completed successfully ...") stay at INFO.
• cartography/intel/gcp/vertex/utils.py, cartography/intel/gcp/cloudrun/util.py — per (resource × region × project) "Found ..." → DEBUG (per-project recap logs stay INFO).
• cartography/intel/aws/sagemaker/util.py — per-(submodule × region × account) "Syncing ..." → DEBUG.
• Sweep: across AWS Bedrock, ECR, CloudFront, IdentityCenter, GitHub, GitLab, Microsoft Entra, Spacelift, SentinelOne, Duo, Ubuntu, Workday, WorkOS, Semgrep, JumpCloud, CVE EPSS, and cartography/util.py, all Fetching/Fetched/Retrieved/Getting/Transforming step logs → DEBUG.
• docs/root/dev/writing-intel-modules.md — removed the redundant "Loading EMR clusters ... into graph" line from the load_emr_clusters example so docs reflect the convention.

Related issues or links

How was this tested?

• Syntax check across the whole cartography/ tree (python3 -c \"import ast, pathlib; [ast.parse(p.read_text()) for p in pathlib.Path('cartography').rglob('*.py')]\").
• make lint locally.
• Manual review of the diff: every demoted call is a per-region / per-tenant step log; every remaining logger.info is a sync-level start/complete or per-account recap message.

Checklist

General

• I have read the contributing guidelines.
• The linter passes locally (make lint).
• I have added/updated tests that prove my fix is effective or my feature works.

Proof of functionality

• Screenshot showing the graph before and after changes.
• New or updated unit/integration tests.

This PR only changes log levels and removes redundant log statements; no graph behavior changes, so no new tests or screenshots are warranted.

Notes for reviewers

• Per-account aggregate INFO logs are intentionally preserved (e.g. CloudTrail Syncing N regions with X hour lookback period, ... sync completed successfully ...).
• cartography/intel/aws/emr.py:80 still has a "Loading EMR ... into graph" line equivalent to the ones removed from DynamoDB; left intact in this PR to keep scope tight, but it could be removed as a follow-up since client/core/tx.py:830 already logs the same information.

[cubic-dev-ai]

No issues found across 50 files

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.