Reduce amount of used `ets` tables

[Strech]

In the current situation, the number of ets tables is proportional to the number of schemas. Every time we parse nested schema we create a new ets schema that goes into recursion.

Despite that fact we already have a cache on top of the schema resolution which is a single place for all schemas after they are resolved we don't leverage it.

It will allow us to reduce coding in related issue #115

TODO

• Measure current usage of the ets tables #124
• Reduce the number, preferably, to a single table
• OPTIONAL: Eliminate the need for a cache or combine a single table and cache