Adding Benchmarks back into agbench

[husseinmozannar]

Why are these changes needed?

Added benchmarks back into autogen, notably GAIA, AssistantBench, HumanEval and WebArena. Renamed everything to be consistent with new package, enabled creation of endpoint from environment file.

Modified readme of magentic-one with a few changes regarding installation instructions

Fixed renaming issues in agbench to be consistent with new package structure.

Checks

• [ x] ran AssistantBench, GAIA and HumanEval on a few examples, but not entirely. So far behavior as expected.
• [] no testing for WebArena so far

TO-DO

• formatting checks and others
• double check from magentic-one team

[husseinmozannar]

Update: reverted all changes outside of agbench and magentic-one readme only.

Might have renamed things outside of my reach by mistake, can revert those (but most seem to make sense)

[husseinmozannar]

Will continue working on agbench for a while. The documentation is outdated, and there are some important features that are necessary which I am still implementing.