modal ci: fix group concurrency (#7691)
currently only one modal CI job is possible across all PRs, which is not
workable - all running jobs get cancelled on a new PR or existing PR
update - fixing the dependency to make the group concurrency work across
PRs not to waste valuable resources.
Signed-off-by: Stas Bekman <stas.bekman@snowflake.com>
Co-authored-by: Stas Bekman <stas.bekman@snowflake.com>
