Refactor smithy-rules-engine

[kstich]

Replaces #1821

---

Significant refactoring of smithy-rules-engine

This commit collects many changes to limit the public interface,
align with Java standards, align with idioms from the rest of the
Smithy codebase, and generally clean up the codebase.

Several packages, classes, and methods have been removed, renamed,
moved, or changed visibility. Use of assert has been removed and
the usage of stream has been reduced. Classes with multiple subtypes
have had their subtype names adjusted for compatibility with Java and
been broken out to their own packages and classes.

---

Enable RuleSet and Test validation by default

This commit removes the custom validation system that required code
to validate a ruleset and its tests, integrating the validation with
Smithy's built in ModelValidation system. Several bugs in validators
and test cases were fixed. Test cases were moved to using the standard
errorfiles based setup where applicable.

Some related refactoring was also performed.

---

Add validation of endpoint test operation inputs

This commit adds validation that for every endpoint test case that
contains operationInputs, the operation input entries match their
specified operation's input structure.

Co-authored-by: Russell Cohen [email protected]

---

Continued refactoring of smithy-rules-engine
This commit collects many changes to limit the public interface,
align with Java standards, align with idioms from the rest of the
Smithy codebase, and generally clean up the codebase.

Several packages, classes, and methods have been removed, renamed,
moved, or changed visibility. Direct access to inner mutable types
has been removed. The internal complexity of the CoverageChecker
has been significantly reduced while maintaining used functionality.

---

Refactor rules engine functions

This commit refactors the configuration and inheritance hierarchy
of functions in the rules engine. Two intermediate abstract classes
have been eliminated, SingleArgFunction and Function.

All functions are now supplied through an SPI on FunctionDefinition,
and each implementation now supplies a FunctionDefinition. Special
cases for methods when generating from FunctionNode instances have
also been removed, aligning all implementations to the same
construction. No behavior has changed in the ExpressionVisitor to
reduce the impact of migration.

Unused, or minimally test-case used, convenience functions have been
removed.

Several bugs were fixed, including dropping context information from
certain sets of RulesError generation.

---

Extract AWS specific rules engine functions

This commit pulls the AWS specific rules engine functions into a new
package, smithy-aws-rules-engine. This includes parseArn, partition,
and isVirtualHostableS3Bucket. The partition loading has also moved
to this package, including its data types.

Relevant tests were migrated. Behavior that was only possible to
validate using the parseArn function have been moved, even if the
behavior is not AWS specific. Tests that used AWS functions but did
not require it were updated to be agnostic.

---

Update license headers

---

Add javadoc and clean up public interfaces

---

Address CR Feedback

---

Extract AWS built-ins and auth validation

This commit updates the endpoint rule set SPI to provide several
components to the implementation, built-in parameters and auth
scheme validators in addition to library functions. AWS specific
components have been extracted to their own package, and core
components are provided through their own extension.

---

Address more CR feedback

---

Add rules-engine specifications

---

Improve in-code function composition

---

Allow disableNormalizePath for sigv4/sigv4a

Add the disableNormalizePath property as an allowed property for
sigv4 and sigv4a auth schemes. This parameter is required for some
services (eg: S3) to correctly sign requests. Specifying it in
endpoint parameters will allow SDKs to potentially remove more S3
customizations.

Co-authored-by: Alex Woods [email protected]

---

Add ModelTransformerPlugin for endpoint test cases

This commit adds a ModelTransformerPlugin implementation that removes
test case entries that rely on operations being removed from a model.
If all test cases are removed, the trait is also removed.

---

Refactor EndpointRuleSetExtension SPI loading

Updates the loading of the EndpointRuleSetExtension SPI to use a lazy
instance holder that is loaded when referenced instead of a lock-based
static initializer. The APIs that access the holder are labeled internal,
as the access patterns may change if the content needs to vary by
instance of the EndpointRuleSet.

---

Move AWS content to smithy-aws-endpoints

---

Fix issues validating rule/test params

---

Add ImplicitGlobalRegion to Partition output

---

Improve in-code rule-set composition

This commit adds a SyntaxElement abstract class that contains several
methods intended for use in constructing rule-sets in code. This class
implements two new interfaces (ToCondition and ToExpression) to support
converting between types of syntax elements that compose into rule-sets.
The Condition, Parameter, and Expression classes extend this,
allowing all types of ruleset syntax to be composed this way.

---

Remove deprecated NumberNode method use

---

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.