Add batch mode to QueryPipeline

[xpbowler]

Description

Add batch mode to QueryPipeline.

Added:

batch parameter to run, arun, _run, _arun, run_multi, arun_multi`

run_multi/arun_multi: Accepts inputs wrapped around in a List, List(Any) rather than List

• New helper function merge_dicts that re-formats the batch output of these 2 functions
• Uses list comprehension to generate individual inputs to _run_multi() (same as single input mode) from the batch input.

_run/_arun: Accepts inputs wrapped around in a List.

• Uses list comprehension to generate individual inputs to _run_multi() (same as single input mode) from the batch input.

I didn't create a batch mode for the with_intermediates functions for now because I don't think its very useful. The list comprehension functions might be confusing tho. I think it might need some more work.

Type of Change

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

[xpbowler]

@logan-markewich changed functions to run batch concurrently with async and validate that size of batch inputs are all the same.

[logan-markewich]

@xpbowler Ok, I think this is in decent shape! Do you want to add an example in the module_guides docs somewhere?

[xpbowler]

@xpbowler Ok, I think this is in decent shape! Do you want to add an example in the module_guides docs somewhere?

updated docs! lmk how it is

[xpbowler]

@logan-markewich hmm, some of the llama-index-integrations failed the unit test. I only changed the .md documentation since last time - did some requirements change?

[logan-markewich]

@xpbowler ah, there was some nasty CICD issues that I fixed on Monday. I'll merge main again and see if it works