Clamp all integer columns

[giancarloaf]

Fixes #52

Validation

• unit testing
• local testing on data under 64-bit
• @rviscomi dataflow test with data over 64-bit

[rviscomi]

Ran this query to find a few test cases with known big ints:

WITH req AS (
  SELECT
    pageid,
    LENGTH(REGEXP_EXTRACT(resp_cache_control, r'(?i)max-age\s*=\s*(\d+)')) AS cc_digits
  FROM
    `httparchive.summary_requests.2022_04_01_mobile`
), pages AS (
  SELECT
    pageid,
    wptid
  FROM
    `httparchive.summary_pages.2022_04_01_mobile`
)

SELECT DISTINCT
  wptid,
  cc_digits
FROM
  req
JOIN
  pages
USING
  (pageid)
WHERE
  cc_digits > 20
ORDER BY
  cc_digits DESC
LIMIT
  5

Ran android-Apr_1_2022/220410_MxAZ_NEMA.har.gz through the pipeline synced with this branch, and it completed successfully.