MER Strategy + OnlineSupervisedMetaLearning Template

[HamedHemati]

This PR adds:

• OnlineSupervisedMetaLearningTemplate: added to the list of common templates
• MER Strategy: an online meta-learning-based strategy (https://arxiv.org/abs/1810.11910)

The algorithm is adapted from version 1 of the algorithm in the official repository:
https://github.com/mattriemer/MER/blob/master/model/meralg1.py

Unfortunately, the results are only available for MNIST and Omniglot in the original paper, and the rest are RL experiments. In the first run, I tested it SimpleMLP from Avalanche on PMNIST, and the results were a little bit higher compared to what is reported in the paper (~0.76 vs. ~0.73), which I guess should be related to the higher capacity of the default MLP model in Avalanche. I can also run it for the original architecture, but since the algorithm is super slow due to the multiple levels of meta-updates in each iteration, I am going to add them later.

[coveralls]

Pull Request Test Coverage Report for Build 3569271524

• 12 of 17 (70.59%) changed or added relevant lines in 4 files are covered.
• 4 unchanged lines in 2 files lost coverage.
• Overall coverage increased (+0.003%) to 73.734%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
avalanche/training/templates/common_templates.py	8	9	88.89%
avalanche/logging/interactive_logging.py	1	5	20.0%

Files with Coverage Reduction	New Missed Lines	%
avalanche/benchmarks/scenarios/classification_scenario.py	2	86.31%
avalanche/training/templates/base.py	2	91.06%

Totals
Change from base Build 3541326024:	0.003%
Covered Lines:	13646
Relevant Lines:	18507

---

💛 - Coveralls

[AntonioCarta]

Should be SupervisedPlugin

[HamedHemati]

Fixed for all!

[AntonioCarta]

For the reproducibility we can discuss this on the continual-learning-baselines repo. If we can have the original architecture it's better.