Data fetcher instance optimization in GQL API

[lukashornych]

Lots of non-parameterized data fetcher are instantiated for each usage. This is not very optimal for memory usage. All usages of such data fetcher should share a single instance of that data fetcher.

Other parameterized data fetchers could be maybe optimized to a single shared instance as well with help of GQL local context passed through data fetchers. We should look into that.

• optimize non-parameterized data fetcher usage
• look into optimizing parameterized data fetcher usage

[lukashornych]

Optimized non-parameterized data fetcher usages are implemented in 677-entity-archivation.

Regarding the parameterized data fetchers, after some examination, there doesn't seem to be an easy way of sharing instances, because in such case we don't have any other way of passing parameters from schema to a data fetcher. Trying to pass the parameters into GQL local context is usually too late.