SPARKNLP-835: ProtectedParam and ProtectedFeature

[DevinTDHa]

Description

This PR enables Params and Features to be protected. Once set, they can't be changed again. This is useful for specific pretrained models, where some parameters should not be changed (as it could interfere with the functionality of the pretrained model).

Protected Features

Protected Features were already introduced in the PR #13777 but are finalized in this PR. An warning will be printed when trying to set a protected feature and it will not be set.

Protected Params

Protected Params are introduced with the trait HasProtectedParams:

https://github.com/DevinTDHa/spark-nlp/blob/0fdab1d0c6eb32125558b77f5d7e33adcbcd5346/src/main/scala/com/johnsnowlabs/nlp/HasProtectedParams.scala

It introduces an implicit conversion from a Param to a ProtectedParam. This can be done, by calling setProtected() on a regular Spark parameter. For Example:

  val maxSentenceLength =
    new IntParam(this, "maxSentenceLength", "Max sentence length to process").setProtected()

The conversion enables additional functionality while setting the value for the parameter. For the set function, it will check, whether or not the (now) ProtectedParam is protected. If so, it will only allow for the parameter to be set once:

spark-nlp/src/main/scala/com/johnsnowlabs/nlp/HasProtectedParams.scala

Lines 47 to 56 in 0fdab1d

	def set[T](param: ProtectedParam[T], value: T): this.type = {
	if (param.isProtected && get(param).isDefined)
	println(
	s"Warning: The parameter ${param.name} is protected and can only be set once." +
	" For a pretrained model, this was done during the initialization process." +
	" If you are trying to train your own model, please check the documentation.")
	else
	set(param.toParam, value)
	this
	}

Note: Default values do not count as set and can therefore be overridden

Comment:
@maziyarpanahi I would like to hear your opinion on this! Do you think this is ok to protect regular Spark parameters?
I wrote this trait, in hopes that we can reuse it often and don't have to change much of the existing code. An example of what kind of code changes this would require, can be seen in the class BertSentenceEmbeddings

Motivation and Context

Some parameters should not be changed after initialization. This PR addresses this missing feature.

How Has This Been Tested?

Tests have been added. The behavior can be seen here:

spark-nlp/src/test/scala/com/johnsnowlabs/nlp/HasProtectedParamsTestSpec.scala

Lines 30 to 41 in 0fdab1d

	it should "set protected params only once" taggedAs FastTest in {
	model.setProtectedParam("first")
	assert(model.getProtectedParam == "first")
	val output = captureOutput {
	model.setProtectedParam("second")
	}
	assert(output.contains("is protected and can only be set once"))
	assert(model.getProtectedParam == "first")
	}

TODO:

• Some tests are failing, ProtectedParams will need to override some additional functions.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• Code improvements with no or little impact
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have read the CONTRIBUTING page.
• I have added tests to cover my changes.
• All new and existing tests passed.

[coveralls]

Pull Request Test Coverage Report for Build 4992350528

• 17 of 24 (70.83%) changed or added relevant lines in 6 files are covered.
• 52 unchanged lines in 34 files lost coverage.
• Overall coverage decreased (-0.01%) to 65.831%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/main/scala/com/johnsnowlabs/nlp/annotators/audio/Wav2Vec2ForCTC.scala	0	1	0.0%
src/main/scala/com/johnsnowlabs/nlp/annotators/coref/SpanBertCorefModel.scala	0	1	0.0%
src/main/scala/com/johnsnowlabs/nlp/annotators/cv/ViTForImageClassification.scala	0	1	0.0%
src/main/scala/com/johnsnowlabs/nlp/annotators/ld/dl/LanguageDetectorDL.scala	2	4	50.0%
src/main/scala/com/johnsnowlabs/nlp/serialization/Feature.scala	6	8	75.0%

Files with Coverage Reduction	New Missed Lines	%
src/main/scala/com/johnsnowlabs/nlp/annotators/DateMatcher.scala	1	97.2%
src/main/scala/com/johnsnowlabs/nlp/annotators/DateMatcherUtils.scala	1	87.5%
src/main/scala/com/johnsnowlabs/nlp/annotators/er/EntityRulerModel.scala	1	90.4%
src/main/scala/com/johnsnowlabs/nlp/annotators/GraphExtraction.scala	1	78.1%
src/main/scala/com/johnsnowlabs/nlp/annotators/ld/dl/LanguageDetectorDL.scala	1	66.27%
src/main/scala/com/johnsnowlabs/nlp/annotators/ner/crf/NerCrfModel.scala	1	69.23%
src/main/scala/com/johnsnowlabs/nlp/annotators/ner/dl/NerDLApproach.scala	1	81.03%
src/main/scala/com/johnsnowlabs/nlp/annotators/ner/dl/NerDLModel.scala	1	56.96%
src/main/scala/com/johnsnowlabs/nlp/annotators/NormalizerModel.scala	1	71.43%
src/main/scala/com/johnsnowlabs/nlp/annotators/parser/dep/GreedyTransition/DependencyMaker.scala	1	80.45%

Totals
Change from base Build 4945586143:	-0.01%
Covered Lines:	8643
Relevant Lines:	13129

---

💛 - Coveralls