Follow-up: Harden RunQuery timestamp/date coercion

[jwildfire]

This Issue was drafted by GitHub Copilot using GPT-5.3-Codex and reviewed by Jeremy Wildfire (@jwildfire)

Task Summary

Track and implement a safe coercion path for timestamp/date parsing in RunQuery() when bUseSchema = TRUE.

Problem / Rationale

A proposed PR update suggested replacing purrr::list_c(parsed) %>% as.POSIXct(...) with an explicit vector conversion via unlist(..., use.names = FALSE) before as.POSIXct(...).

That change was intentionally deferred from PR #39 to keep the release candidate focused. The concern remains valid and should be addressed in a targeted follow-up with explicit regression coverage.

Scope

• Review current parsing behavior in R/RunQuery.R for schema-based Date/timestamp coercion.
• Confirm edge cases for list-like parsed values and mixed successful/failed conversions.
• Implement robust vector coercion before POSIXct conversion if needed.
• Add/update tests in tests/testthat/test-RunQuery.R for this exact parsing path.

Out of scope:

• Broader RunQuery refactors unrelated to date/timestamp coercion.
• API/interface changes to RunQuery().

Proposed Approach (Optional)

• Reproduce current behavior with representative inputs (character timestamps, invalid strings, NA values).
• Update conversion to use explicit vector flattening where appropriate.
• Preserve warning/logging behavior for unparsable values.
• Validate with focused tests and devtools::check().

QC Approach

In addition to standard QC (e.g., code reviews, automated checks) the following QC measures will be implemented:

[ ] Qualification Test via Double Programming
[x] Unit Tests
[ ] User Tests (e.g. visual comparison)