Benchmarks performance improvements

[lrzpellegrini]

This PR brings some improvements on the benchmarks creation side. This improvement mainly concerns the instantiation of AvalancheDatasets.

When profiled using PyCharm, the creation and iteration of a PermutedMNIST scenario with 10 experiences is 10x faster. These optimizations may introduce a small increase in the memory overhead.

[coveralls]

Pull Request Test Coverage Report for Build 726119201

• 9 of 9 (100.0%) changed or added relevant lines in 2 files are covered.
• 4 unchanged lines in 2 files lost coverage.
• Overall coverage decreased (-0.06%) to 76.16%

---

Files with Coverage Reduction	New Missed Lines	%
avalanche/benchmarks/utils/avalanche_dataset.py	1	86.13%
avalanche/benchmarks/utils/dataset_utils.py	3	89.16%

Totals
Change from base Build 725213338:	-0.06%
Covered Lines:	6453
Relevant Lines:	8473

---

💛 - Coveralls