Implement Scheduling Optimizations and Enhance Dependency Resolution and Benchmarking Mechanisms

[gliargovas]

In this extensive pull request, major improvements and refinements have been carried out (beyond the initial focus on benchmarks and reporting).

Dependency Resolution

• Introduced an abstraction over early and late resolution of dependencies, allowing for more accurate tracking of file system and environment changes.
• Refactored the environment resolution process to improve the code quality and to accommodate the new early resolution mechanism.

Benchmarking Enhancements

• Added a newscript for running and tracking the performance benchmarks and included additional scripts for expanded benchmarking.
• Enhanced the benchmark execution reporting with Gantt chart representation and execution statistics including command execution counts and time metrics.

Scheduler Optimizations

• Updated several operational components within the scheduler to accommodate the new early dependency resolution and to improve the scheduler's overall functionality.
• Added optional arguments for scheduler optimization and enhanced the logging with timestamped entries for better traceability.

Miscellaneous Refinements:

• Fixed various minor bugs, typos, and correctness issues.
• Reimplemented process killing using psutil for better management of process termination.
• Added support for more complex timestamp logging.

Code and Documentation Quality:

• Conducted minor refactoring across the codebase to improve the readability and maintainability.
• Expanded the README documentation for better guideline on tests, benchmarks and the top-level directory.

[gliargovas]

I manually ran the tests, and everything passed (the error was in comment on PR) - with or without sandbox killing

[angelhof]

Please change the title and description to reflect the actual scope of this PR.

Feel free to merge without additional approval after you resolve my comments :)

[angelhof]

Add a comment here that this won't work if we move to a full partial order, ideally together with an assertion that len(predecessor) <=1 so that it fails once we move to a proper full partial order