added golden USCensusPEP_PopulationEstimatebyRace

[niveditasing]

This PR includes:

1. Robust Download Retries & Timeout Handling ( urlllib3 & Session integration )
   To prevent intermittent run-time connection failures or GCS/Census endpoint timeouts:
   • Added Urllib3 adaptors: Integrated standard Retry and HTTPAdapter at the module level.
   • Persistent Session: Replaced individual requests.get calls with a shared requests.Session() object configured to automatically retry
     failed requests (up to 5 retries, handling status codes 429, 500, 502, 503, 504 with exponential backoff).
   • Passed Session to downstreams: Forwarded the configured session object to downstream loaders like _resolve_pe_11 so they benefit from
     the same robust timeout and retry safety.

---

2. Safe Coordinate-Based DataFrame Access (iloc Refactoring)
   To eliminate risk of unexpected behavior, performance side-effects, and standard Pandas warnings (such as SettingWithCopyWarning):

• Replaced chained indexing: Converted chained lookups like df.iloc[j]["Area"] = ... to standard, safe coordinate-based syntax:
  df.iloc[j, 0] = ....
• Added explicit casting: Wrapped coordinates inside explicit str(...) casting to guarantee concatenation safety regardless of the
  column's underlying data type.

---

3. Correction of Length Evaluation Bug
   Fixed a critical logical/syntactical bug in checking filename lengths:

• Before: len(file_to_dowload['file_name'] > 5) (This incorrectly evaluated the length of a boolean comparison expression, causing
  logical failures).
• After: len(file_to_dowload['file_name']) > 5 (Correctly measures the length of the string value).

---

4. Deterministic Input File Scanning (sorted())

• Wrapped os.listdir() in sorted(...) when populating input files list.
• This ensures that input files are read in a deterministic order on any operating system, preventing unpredictable schema mapping,
  mismatched generation, or tests behaving differently on different filesystems.

---

5. Pandas Deprecation Upgrades

• Modernized deprecated parameters under pd.read_table(...).
• Replaced the deprecated delim_whitespace=True with explicit standard separator regex: sep=r'\s+', ensuring perfect compatibility with
  recent and future Pandas versions.

---

6. Added goldens with 0.1% increase in threshold

PR checklist : https://docs.google.com/spreadsheets/d/1BzweR9Sj58j0H2_BweGTmfE4Z1lrjPZL8u1FS0kzCeg/edit?pli=1&gid=0#gid=0

[gemini-code-assist]

Code Review

This pull request adds several golden data CSV files and introduces a validation configuration file (validation_config.json) along with its reference in the manifest to enable automated validation checks for the US Census PEP population estimate by race import. Feedback on the changes highlights a critical filename mismatch in the validation configuration that would cause a file-not-found error, as well as recommendations to ensure all validation rules have unique rule_id values for clearer reporting.

[gemini-code-assist]

The refactoring of iloc indexing and the fix for the length evaluation bug are important improvements for data integrity and performance. Using sorted(os.listdir()) is a good practice for ensuring deterministic behavior across different filesystems. The transition to requests.Session with urllib3 retries will significantly improve the robustness of the data ingestion process. One suggestion: verify that the validation_config.json thresholds are sufficiently strict to catch regressions in the golden files, as the 0.1% threshold should be evaluated against the expected variance of the Census data.

[niveditasing]

@gemini-code-assist review

[gemini-code-assist]

Code Review

This pull request introduces validation configurations and golden summary reports for US Census PEP population estimates by race. It updates preprocess.py to use a robust requests.Session with retry logic, fixes a bug in a string length check, sorts input files, and updates pandas operations to avoid deprecation warnings. Feedback is provided to extend the retry adapter configuration to cover both HTTP and HTTPS protocols.

[gemini-code-assist]

Currently, the retry adapter is only mounted to https:// URLs. If any input URLs or future URLs use http://, they will not benefit from the configured retry mechanism. It is safer and more robust to mount the adapter to both http:// and https:// protocols.

Suggested change

	retries = Retry(total=5,
	backoff_factor=1,
	status_forcelist=[429, 500, 502, 503, 504],
	raise_on_status=False)
	session.mount('https://', HTTPAdapter(max_retries=retries))
	retries = Retry(total=5,
	backoff_factor=1,
	status_forcelist=[429, 500, 502, 503, 504],
	raise_on_status=False)
	adapter = HTTPAdapter(max_retries=retries)
	session.mount('http://', adapter)
	session.mount('https://', adapter)