Set model.eval() in TorchEstimator before return

[thuningxu]

Checklist before submitting

• Did you read the contributor guide?
• Did you update the docs?
• Did you write any tests to validate this change?
• Did you update the CHANGELOG, if this change affects users?

Description

TorchEstimator currently returns a model without switching to eval mode. This PR fixes that.

Review process to land

1. All tests and other checks must succeed.
2. At least one member of the technical steering committee must review and approve.
3. If any member of the technical steering committee requests changes, they must be addressed.

[github-actions]

Unit Test Results

   465 files  +   179     465 suites  +179   4h 4m 41s ⏱️ + 1h 36m 45s
   533 tests ±       0     504 ✔️ +     38       29 💤  -   36  0 ❌  - 2 
9 398 runs  +3 530  7 517 ✔️ +3 058  1 881 💤 +476  0 ❌  - 4 

Results for commit 2fdea15. ± Comparison against base commit e94d8ea.