Fix profiler overwrite table semantics

[dgomez04]

Issue

Profiler overwrite mode replaced rows with TRUNCATE + INSERT while retaining the existing DuckDB table definition. For tables created through inferred schemas, a first run could persist a narrow type such as DECIMAL(10,8). A same-day rerun with larger values then failed during insertion because the stale schema was reused.

Solution

• Replace inferred-schema tables with DROP + CREATE TABLE AS SELECT on overwrite so both the definition and data reflect the current run.
• Recreate DDL-owned tables before loading them so explicitly declared types remain authoritative.
• Align the Python profiler DuckDB helper with the same replace-table semantics.
• Update the existing overwrite tests to cover schema replacement and narrow-to-wide Decimal values.

Tests: .venv/bin/python -m pytest tests/unit/assessment/test_duckdb_helpers.py -q --tb=short (13 passed)

Follow-up

We need to consolidate profiler writes behind a single API. _save_overwrite, save_to_duckdb, and _save_to_db currently provide overlapping write paths and can drift in behavior. A separate issue will be created to track that consolidation.

[codecov]

Codecov Report

❌ Patch coverage is 8.69565% with 21 lines in your changes missing coverage. Please review.
✅ Project coverage is 70.37%. Comparing base (25225cb) to head (6198245).

Files with missing lines	Patch %	Lines
...databricks/labs/lakebridge/assessments/pipeline.py	0.00%	21 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2581      +/-   ##
==========================================
- Coverage   70.40%   70.37%   -0.04%     
==========================================
  Files         107      107              
  Lines        9661     9661              
  Branches     1070     1068       -2     
==========================================
- Hits         6802     6799       -3     
- Misses       2653     2657       +4     
+ Partials      206      205       -1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

✅ 174/174 passed, 9 flaky, 2 skipped, 1h9m37s total

Flaky tests:

• 🤪 test_installs_and_runs_local_bladebridge (13.431s)
• 🤪 test_installs_and_runs_pypi_bladebridge (22.58s)
• 🤪 test_recon_for_report_type_is_data (37.18s)
• 🤪 test_transpiles_informatica_to_sparksql_non_interactive[True] (21.822s)
• 🤪 test_transpiles_informatica_to_sparksql (24.142s)
• 🤪 test_transpiles_informatica_to_sparksql_non_interactive[False] (4.084s)
• 🤪 test_transpile_teradata_sql_non_interactive[True] (5.756s)
• 🤪 test_transpile_teradata_sql (27.104s)
• 🤪 test_transpile_teradata_sql_non_interactive[False] (22.911s)

_{Running from acceptance #5138}