Re-implement TIR example

[ipetr0v]

After a discussion in #1066 we need to re-implement Trusted Information Retrieval example in the following way:

• A separate TIR node should be created for each new client request (similar to Aggregator example should not create new nodes from existing nodes #1246)
• database_proxy node should download the whole external database
  • We are removing a memory restriction and start assuming that the whole database could fit into Oak
• TIR node shouldn't send any requests to the database_proxy node, instead we need a router node that:
  • Will not be tainted with user data (it just forwards invocations)
  • Will request the database and pass it to the TIR node

The idea behind these changes is to remove the necessity of trusting any particular node in the application and just to assign a client token label to data.

cc @tiziano88 @conradgrobler