A brief clarification. Google uses a controlled vocabulary defined by
schema.org. Publishers may present information using schema.org
terms. If a Publisher uses terms from schema.org, it also must structure the information to conform to the schema.org
grammar. For example, for your
Reviews, terms for
AggregateRating must be allowable and complete (Google requires specific data). When a Publisher presents structured information, it may expose the data to humans on a web page using
Microdata (Rich Snippets) or
RDFa. If the Publisher has correctly used the schema.org terms and grammar, the Google Crawler will parse and correctly interpret the information.
Publisher also may present information in a machine-readable format. That format is
JSON-LD. The JSON-LD structure could package a collection of Reviews on a single page. Publisher will want to make sure that the meaning of the JSON-LD structure has the same meaning as the HTML markup (either Microdata or RDFa). Google may marginalize a site where human-readable data is not semantically equivalent to machine-readable data.
You ask about "extra features" for pagination. The JSON-LD can be packaged as a single structure to describe all semantic information on a page. That structure may be anywhere in the HTML document. If you chunk your Reviews in some order on the page, the RDFa or Microdata will be associated with the chunks. Otherwise, I'm not use what you mean by pagination. You ask: "Does Google Crawler work for all reviews on a page?" The answer is yes.