Xcsf

Changes:
* Remove compilation with `-flto` ([112](https://github.com/xcsf-dev/xcsf/pull/128))

Changes:
* Improve parameter checks ([112](https://github.com/xcsf-dev/xcsf/pull/112), [#114](https://github.com/xcsf-dev/xcsf/pull/114))
* Improve hyperrectangle CSR semantics ([117](https://github.com/xcsf-dev/xcsf/pull/117))
* Fix a sampling bug in Python `fit()` when `shuffle=False` ([122](https://github.com/xcsf-dev/xcsf/pull/122))

Changes:
* Make predict function deterministic even with parallelism ([109](https://github.com/xcsf-dev/xcsf/pull/109))

Changes:
* Fix a bug serialising initial population set filename ([104](https://github.com/xcsf-dev/xcsf/pull/104))
* Add tests ([105](https://github.com/xcsf-dev/xcsf/pull/105))

Changes:
* Make `random_state` parameter sklearn compatible ([97](https://github.com/xcsf-dev/xcsf/pull/97))

Changes:
* Add tests.
* Fix a bug serialising initial population set filename.
* Fix a bug copying convolutional layer momentum.