Fix more intermittent tests in Test_Client_JobCompletion

[brandur]

Fix a series of tests in Test_Client_JobCompletion that may fail [1]
intermittently due to their use of a real world clock that may be off as
test goroutines are paused in GitHub Actions for more extended periods
of time.

--- FAIL: Test_Client_JobCompletion (0.00s)
    --- FAIL: Test_Client_JobCompletion/JobThatReturnsJobCancelErrorIsImmediatelyCancelled (4.81s)
        client_test.go:5586: Generated schema "river_2025_04_26t20_41_17_schema_18" with migrations [1 2 3 4 5 6] on line "main" in 2.546342002s [18 generated] [14 reused]
        logger.go:256: time=2025-04-26T20:41:23.623Z level=INFO msg="River client started" client_id=fv-az1370-305_2025_04_26T20_41_23_606599
        client_test.go:5600:
                Error Trace:	/home/runner/work/river/river/client_test.go:5600
                Error:      	Max difference between 2025-04-26 20:41:25.839484186 +0000 UTC m=+8.375195240 and 2025-04-26 20:41:23.819235 +0000 UTC allowed is 2s, but difference was 2.020249186s
                Test:       	Test_Client_JobCompletion/JobThatReturnsJobCancelErrorIsImmediatelyCancelled
        logger.go:256: time=2025-04-26T20:41:25.862Z level=INFO msg="River client stopped" client_id=fv-az1370-305_2025_04_26T20_41_23_606599
        riverdbtest.go:277: Checked in schema "river_2025_04_26t20_41_17_schema_18"; 1 idle schema(s) [18 generated] [15 reused]
FAIL
FAIL	github.com/riverqueue/river	32.869s

Fix these by injecting a test clock and asserting time based on that
instead of the real wall clock.

[1] https://github.com/riverqueue/river/actions/runs/14684878766/job/41212306331

[bgentry]

🤘 thanks