A variety of system test targets fail running loads with java.lang.Math.tanh

See March 2025 release triage https://github.com/adoptium/aqa-tests/issues/6051 for many instances of system test targets failing across many platforms with similar or same issues (`FAIL: gnu.testlet.java.lang.Math.tanh (number 4) got 4374121138083594055 but expected 4374121138083594054` errors)

```
[2025-03-19T08:15:03.744Z] LT 08:15:02.373 - Test failed
[2025-03-19T08:15:03.744Z] LT Failure num. = 1
[2025-03-19T08:15:03.744Z] LT Test number = 1959
[2025-03-19T08:15:03.744Z] LT Test details = 'Mauve[gnu.testlet.java.lang.Math.tanh]'
[2025-03-19T08:15:03.744Z] LT Suite number = 0
[2025-03-19T08:15:03.744Z] LT Thread number = 3
[2025-03-19T08:15:03.744Z] LT >>> Captured test output >>>
[2025-03-19T08:15:03.744Z] LT PASS: gnu.testlet.java.lang.Math.tanh (number 0)
[2025-03-19T08:15:03.744Z] LT PASS: gnu.testlet.java.lang.Math.tanh (number 1)
[2025-03-19T08:15:03.744Z] LT PASS: gnu.testlet.java.lang.Math.tanh (number 2)
[2025-03-19T08:15:03.744Z] LT PASS: gnu.testlet.java.lang.Math.tanh (number 3)
[2025-03-19T08:15:03.744Z] LT FAIL: gnu.testlet.java.lang.Math.tanh (number 4)
[2025-03-19T08:15:03.744Z] LT got 4374121138083594055 but expected 4374121138083594054
[2025-03-19T08:15:03.744Z] LT FAIL: gnu.testlet.java.lang.Math.tanh (number 5)
[2025-03-19T08:15:03.744Z] LT got -4849250898771181753 but expected -4849250898771181754
```

List of targets failing in this way:
- MiniMix_5m_0
- MiniMix_5m_1
- MiniMix_10m_0
- MiniMix_10m_1
- MiniMix_aot_5m_0
- MauveSingleThrdLoad_HS_5m_0
- MauveSingleThrdLoad_HS_5m_1
- MauveSingleInvocLoad_HS_5m_0
- MauveSingleInvocLoad_HS_5m_1
- MauveMultiThrdLoad_5m_0
- MauveMultiThrdLoad_5m_1

Did not see this during dryrun because of a different failure (masking these ones).

Start with the March 2025 release triage issue 6051 and reproduce the Mauve gnu.testlet.java.lang.Math.tanh failures in one of the listed system-test targets. Compare the reported values and verify the result across the affected targets; done means these failures are resolved or correctly accounted for without being masked by another failure.

Written by the indexing model from the issue text.