visualisation of benchmark results

[jmatejcz]

Is your feature request related to a problem? Please describe.

Describe the solution you'd like
Upgrade and unify gathering results from benchmarks
Add python script showing different charts

Describe alternatives you've considered

Additional context

[jmatejcz]

• add task categories to results in tool_agent_benchmark - manipulation, spatial, etc..

[jmatejcz]

• how to assign error to subtask in all validators?

[jmatejcz]

• group by complexity of tasks
• extra tool calls used by model/task/subtask ?

[jmatejcz]

• update unit tests to changed validators and subtasks

[jmatejcz]

• restructure new code
• README update

[jmatejcz]

• integrate manipulation o3de results same way - partially will be moved to new issue
• computing summary results after every step, in case early stop

[jmatejcz]

applied here:
#542