WIP: Update for 3.x

[krishnangovindraj]

Must do:

• Update readme
• Ensure build & install from source works.

What is the goal of this PR?

Updates the plugin for TypeDB 3.x. The simplified driver implementation & protocol has also allowed us to move to a different architecture. Major changes include:

Major new 'features'

• A split into %typedb line magic for general administrative commands, and %%typeql cell magic for running queries
• A utility library, intended to hold functions such as conversions to networkx, & interfaces such as visualisable graph builders.
• Sample notebooks to serve as a quick-start - including one graph visualisation.

Subtle UX changes

• We only support one implicit transaction at a time - no more aliases.
• We implicitly store the result of the transaction in the _typeql_resultvariable (as well as returning it to the cell) - no magic parameter to specify the variable name. Just re-assign it in the next cell.

Missing features:

• Various output formats such as JSON
• Transaction & Query options

What are the changes implemented in this PR?

I've taken inspiration from the 2.x version, but also taken the freedom to move to a more opinionated architecture which resembles TypeDB console. Specifically:

• The move towards resembling console commands, and the separation of %typedb and %%typeql magics.
• The re-architecture into SubCommands reflecting each %typedb <sub-command>
  • The intent is organisation, as the console command syntax dictates where the code must be.
• A hacky-parser that reconstructs a query structure on the client side, mainly for use with visualisation
  • Notably, an imminent TypeDB feature will make this redundant, as the query-structure will be returned in the response.

I'm not married to any of these and it should be straightforward to re-implement them if there's demand for it.

[krishnangovindraj]

Please hit the '...' on the top-right and select 'view file' to see the rendered version.

[krishnangovindraj]

Please hit the '...' on the top-right and select 'view file' to see the rendered version.

[james-whiteside]

LGTM apart from:

• As the readme is specially treated on GitHub and in IDEs, it would be nice if the contents of the sample notebook could be displayed there, even if it's just a verbatim copy of the sample notebook as-rendedered.
• It would be nice to have variable interpolation back, which is gone unless I'm mistaken. I also don't really like the idea of setting the result to a top-level namespace variable, which may have unintended behaviour. If you don't like providing the variable name as an argument, perhaps it could be accessed via a last_result class variable of the module instead? Moving values in and out of different modules and extensions is pretty key for Jupyter, so it's important that these features work smoothly.