Search model renames

[heaths]

There are a few more renames to make. But not just renames: the "V2" models should be collapsed with their v1 (no suffix, that is) counterparts and original @odata.type retained for round-tripping. When new ones are created, the "V2" equivalent should always be created because the v1s will cause the service to fail when using the new api-version.

Old name	New name	Notes
EdgeNGramTokenFilterV2	EdgeNGramTokenFilter	Collapse with existing EdgeNGramTokenFilter
KeywordTokenizerV2	KeywordTokenizer	Collapse with existing KeywordTokenizer (will have separate properties for v1 and "V2")
LuceneStandardTokenizerV2	LuceneStandardTokenizer	Collapse with existing LucenseStandardTokenizer
NGramTokenFilterV2	NGramTokenFilter	Collapse with existing NGramTokenFilter
PathHierarchyTokenizerV2	PathHierarchyTokenizer	(There is no v1)
Similarity	SimilarityAlgorithm	Do not rename derivative types
Suggester	SearchSuggester