tensorboard functionality

[TarekAbouChakra]

Key changes:

• Tensorboard Class (tblogger): I've added a new tensorboard class, tblogger, which integrates into neps. This integration allows us to take advantage of TensorBoard's visualization.
• Config Data Extraction: I've integrated the extraction of configuration data from both metahyper and the neps API. This step proves incredibly useful for parallel runs.
• Dependency Update: A new dependency has been introduced to support TensorBoard integration in poetry.
• Example Usage: To facilitate easy adoption, I've included a new example under neps_examples/convenience that demonstrates how to run neps with the tblogger class.

Drawback Consideration:

One potential drawback of the tblogger integration is the creation of multiple tfevent files. This was unaviodable to ensure compatibility with parallel runs.

[TarekAbouChakra]

Changes:

• Removing logging gradients and leaving the user's preference by manual calculation and logging through scalars.
• Fixing the example file, making it easier to read and follow.

Remaining issues and TODO's:

• The issue of generating multiple files throughout the run.
• Finalizing the documentation.

[TarekAbouChakra]

Resolved the comments in the last PR concerning:

• Code maintainability and cleaning
• Couple more warnings and error checks
• Better comments

One new feature:

• Added a tblogger.enable() to remove the double negative problem

[TarekAbouChakra]

Mostly minor cleanups.

1 new addition:

• tblogger.get_status() this returns true when logger is enabled and false when it is disabled to keep track of tblogger's state.

[Neeratyoy]

Could we format the example/tutorial code to follow the styling standards better, such that, it looks better?

[TarekAbouChakra]

Changes only for the tblogger example format:

• Maximum of 80 characters per line.
• BO now does only 5 random samples before sampling from acquistion function.
• Better comments