Backports for v0.8.1 (awslabs#1655)

* loosen RTOL in `test/distribution/test_flows.py` to make `test_flow_invertibility` pass (awslabs#1604)

* loosen RTOL to make `test_flow_invertibility` pass

* fix type

Co-authored-by: Eduard Uffelmann <[email protected]>

* Add batch_size attribute to MQCNNEstimator and MQRNNEstimator (awslabs#1645)

* Fix NegativeBinomial constructor args in NegativeBinomialOutput (torch) (awslabs#1651)

* Fix frequency string conversion from ts format, add test (adapted from awslabs#1652)

Co-authored-by: Eduard <[email protected]>
Co-authored-by: Eduard Uffelmann <[email protected]>
Co-authored-by: StanG <[email protected]>

Backports for v0.7.7 (awslabs#1654)

* Fix frequency metadata bug for lstnet datasets (awslabs#1593)

* Add batch_size attribute to MQCNNEstimator and MQRNNEstimator (awslabs#1645)

* Fix NegativeBinomial constructor args in NegativeBinomialOutput (torch) (awslabs#1651)

Co-authored-by: StanG <[email protected]>

add GitHub workflow for publishing to PyPI for new releases (awslabs#…

…1601) (awslabs#1603)

Co-authored-by: Eduard Uffelmann <[email protected]>

Co-authored-by: Eduard Uffelmann <[email protected]>

Fix serde issue with some distribution output types, add test (awslab…

…s#1543) (awslabs#1563)

Co-authored-by: Lorenzo Stella <[email protected]>

Merges for v0.7.5 (awslabs#1536)

* Train invoke with (awslabs#1530)

* fix ProphetPredictor serialization issue (awslabs#1535)

* make ProphetPredictors init_model Optional to make it serializable

* use toolz.identity instead of None, becuase of style-type-checks

* Add test.

Co-authored-by: Eduard Uffelmann <[email protected]>
Co-authored-by: Jasper Zschiegner <[email protected]>

* Add item_id to r forecast predictors (awslabs#1537)

* Serde: Allow encoding of functions and methods. (awslabs#1519)

* Add support for functools.partial.
* Change type for fqname_for argument.

Co-authored-by: Lorenzo Stella <[email protected]>

* Disable tests on Windows for PRs, fix other workflows (awslabs#1525)

Co-authored-by: Jasper <[email protected]>
Co-authored-by: Eduard Uffelmann <[email protected]>
Co-authored-by: Oliver Borchert <[email protected]>
Co-authored-by: Lorenzo Stella <[email protected]>

Merges for v0.7.4 (awslabs#1523)

* Fix Settings._inject to check if it can provide the value. (awslabs#1501)

* Fix indentation (awslabs#1500)

* Fix anomaly detection example (awslabs#1515)

* Add constant dummy time features to TFT for yearly data (awslabs#1518)

Co-authored-by: Jasper <[email protected]>
Co-authored-by: Pavel Zwerschke <[email protected]>
Co-authored-by: Lorenzo Stella <[email protected]>
Co-authored-by: Oliver Borchert <[email protected]>

Fix get_lags_for_frequency for minute data in DeepVAR (awslabs#1455) (a…

…wslabs#1468)

Co-authored-by: Kashif Rasul <[email protected]>

Merges for v0.7.2 (awslabs#1453)

* Fixes for MXNet 1.8 (awslabs#1403)

* Lock the version for mxnet theme to 0.3.15 (awslabs#1451)

* Fix train-test split data leakage for m4_yearly and wiki-rolling_nips. (awslabs#1445)

* Fix missing import in gluonts.mx.model.GluonEstimator (awslabs#1450)

Co-authored-by: Lorenzo Stella <[email protected]>
Co-authored-by: Leonardo Araneda Freccero <[email protected]>

Merges for v0.6.9 (awslabs#1454)

* Fix train-test split data leakage for m4_yearly and wiki-rolling_nips. (awslabs#1445)

* Lock the version for mxnet theme to 0.3.15 (awslabs#1451)

Co-authored-by: Leonardo Araneda Freccero <[email protected]>
Co-authored-by: Lorenzo Stella <[email protected]>

fix compatibility for pandas < 1.1 in `time_feature/_base.py` (awslab…

…s#1437) (awslabs#1438)

* fix compatibility for pandas < 1.1 in `time_feature/_base.py` 

Co-authored-by: Eduard Uffelmann <[email protected]>