Additional Examples for more complex Algorithms built with AAF

[ckunki]

Inspired by comment 2405464444 in PR #67.

After some discussions with @mkcorneli , I also recognized we need to give some examples for the algorithms you can implement with this framework (not actually implementation in the first place), here are some

• Scalable a priori for frequent item set mining on large tables
• Scalable Clustering, like k-means on large tables
• Scalable Graph Algorithms that go beyond the capabilities of connect by, such as Iterative Map Reduce based PageRank

Furthermore, we need to explain what are the benefits compared to running the queries from a Python UDF via pyexasol

• A Python UDF that connects via pyexasol back to the DB that is running that UDF, will use two different sessions and transactions for the Python UDF and the pyexasol connection. This can lead to consistency issues, meaning that one transaction doesn't see the changes of the other, and to deadlocks when both transaction happens to modify the same database objects.
• In the AAF, all queries that a QueryHandler emits and get run by the query loop in a call to AAF_RUN_QUERY_HANDLER, will run in the same session and transaction (if no explicit commits are emitted). This is equal to Lua Scripting Scripts.