reduce test concurrency to avoid HF rate limiting, test suite parity (#2128)

* reduce test concurrency to avoid HF rate limiting, test suite parity

* make val_set_size smaller to speed up e2e tests

* more retries for pytest fixture downloads

* val_set_size was too small

* move retry_on_request_exceptions to data utils and add retry strategy

* pre-download ultrafeedback as a test fixture

* refactor download retry into it's own fn

* don't import from data utils

* use retry mechanism now for fixtures

tests/e2e/patched/test_fused_llama.py

@@ -38,7 +38,7 @@ class TestFusedLlama(unittest.TestCase):
                 "flash_attn_fuse_mlp": True,
                 "sample_packing": True,
                 "sequence_len": 1024,
-                "val_set_size": 0.1,
+                "val_set_size": 0.02,
                 "special_tokens": {
                     "unk_token": "<unk>",
                     "bos_token": "<s>",