Extend accuracy tests for models that we support

[AnetaKaczynska]

This PR extends dev Nightly test suite with accuracy tests for broader range of models.

[michalkuligowski]

What is the delta time with new test suites?

[AnetaKaczynska]

Time taken by the new test suites:

name	time
gsm8k_small_g3_tp1_part2	6m 41s
gsm8k_small_g3_tp1_part3	9m 2s
gsm8k_large_g3_tp2_part2	11m 15s
gsm8k_small_g3_tp1_fp8	22m 26s

[michalkuligowski]

@AnetaKaczynska I was rather more interested in the time taken by the test itself, those measurements are for whole suite including environment preparation (which can vary for many reasons, and will vary also for existing suites). The exact time for the suite can be measured by taking the "============================= test session starts ==============================" time and test finish

[AnetaKaczynska]

suite	model	time1	time2
gsm8k_small_g3_tp1_part2	granite-8b.yaml	00:01:21	00:01:24
gsm8k_small_g3_tp1_part2	granite-20b.yaml	00:01:48	00:01:51
gsm8k_small_g3_tp1_part3	Qwen2-7b-Instruct.yaml	00:01:13	00:01:16
gsm8k_small_g3_tp1_part3	Mistral-7B-Instruct-v0.3.yaml	00:01:13	00:01:16
gsm8k_large_g3_tp2_part2	Mixtral-8x7B-Instruct-v0.1.yaml	00:01:39	00:06:29
gsm8k_small_g3_tp1_fp8	granite-8b-fp8.yaml	00:01:23	00:05:42
gsm8k_small_g3_tp1_fp8	granite-20b-fp8.yaml	00:01:51	00:05:38

Ok, for clarity I gathered two values:

1. test time as reported directly in logs (e.g. ================== 1 passed, 3 warnings in 111.16s (0:01:51) ===================)
2. time taken between 'test session starts' and 'PASSED MODEL', which for large and fp8 models is usually several minutes longer.