Fix Snowflake resync failure due to case mismatch in _resync table suffix

[kang8]

Summary

Fixes #4184

• The _resync suffix was appended in lowercase, producing mixed-case identifiers like ACCOUNTS_resync
• getColsFromTable() in qrep.go:187 unconditionally applies strings.ToUpper(), converting it to ACCOUNTS_RESYNC
• Snowflake treats quoted identifiers as case-sensitive, so the table lookup fails with "does not exist"
• Changed the suffix to uppercase _RESYNC to ensure consistent identifiers across creation and query paths

Test plan

• Verified Snowflake resync creates tables with _RESYNC suffix
• Verified getColsFromTable() can find the resync tables
• Verified table swap (TrimSuffix) correctly strips _RESYNC after resync completes
• Confirm no regression for BigQuery, ClickHouse, or PostgreSQL destinations (they don't use getColsFromTable with ToUpper)

[CLAassistant]

All committers have signed the CLA.

[kang8]

Hi @serprex @Amogh-Bharadwaj, friendly ping on this PR — it's a small 2-line fix for a Snowflake resync bug (#4184). CLA is signed and CI is waiting on a maintainer to approve the workflow run. Could one of you take a look when you have a moment? Thanks!

[serprex]

cdc_flow.go is shared code between all connectors, this should be fixed in snowflake connector specifically

[kang8]

Thanks for the review! I've moved the fix to be Snowflake-specific in getColsFromTable (connectors/snowflake/qrep.go) and left cdc_flow.go untouched. PTAL when you have a moment.

[codecov]

❌ 2 Tests Failed:

Tests completed	Failed	Passed	Skipped
2253	2	2251	203

View the top 2 failed test(s) by shortest run time

github.com/PeerDB-io/peerdb/flow/e2e::TestApiMongo

Stack Traces | 0.01s run time

=== RUN   TestApiMongo
=== PAUSE TestApiMongo
=== CONT  TestApiMongo
--- FAIL: TestApiMongo (0.01s)

github.com/PeerDB-io/peerdb/flow/e2e::TestApiMongo/TestClickHouseMirrorValidation_Pass

Stack Traces | 0.06s run time

=== RUN   TestApiMongo/TestClickHouseMirrorValidation_Pass
=== PAUSE TestApiMongo/TestClickHouseMirrorValidation_Pass
=== CONT  TestApiMongo/TestClickHouseMirrorValidation_Pass
    api_test.go:306: 
        	Error Trace:	.../flow/e2e/api_test.go:306
        	Error:      	Received unexpected error:
        	            	rpc error: code = FailedPrecondition desc = failed to validate source connector mongo: oplog retention must be set to >= 24 hours, but got 23.000000
        	Test:       	TestApiMongo/TestClickHouseMirrorValidation_Pass
    api_test.go:49: begin tearing down postgres schema api_3395sqco
--- FAIL: TestApiMongo/TestClickHouseMirrorValidation_Pass (0.06s)

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}

[kang8]

The Codecov failure TestApiMongo/TestClickHouseMirrorValidation_Pass (oplog retention must be set to >= 24 hours, but got 23.000000) is unrelated to this PR — it's a pre-existing flaky test.

Root cause: TestApiMongo runs subtests in parallel via RunSuite (e2eshared.go:24). TestMongoDBOplogRetentionValidation (api_test.go:1138) temporarily mutates the shared MongoDB instance's global oplog retention to 23h to exercise the failure path, then restores it to 24h. Any other parallel subtest that performs CDC validation during this window fails.

Prior occurrences of the same flake (different subtests, same error):

• Fix password encoding in Postgres connection strings #4096 — TestApiMongo/TestCancelTableAddition_NoRemovalAssumed
• improve schema validations gaps #3752 — MongoDB subtest

This PR only touches flow/connectors/snowflake/qrep.go; it does not touch any MongoDB / ClickHouse / validation code paths. Could we rerun the failed job (or merge as-is)? I'm happy to open a separate issue to track fixing the flake.