prevent `cortex deploy` from hanging if users run in directories with large files (e.g. models)

[deliahu]

Description

If possible, it should be more obvious in the docs and/or in the product that the entire contents at and below cortex.yaml gets zipped and sent to the cluster. This will avoid situations such as the user's model being unintentionally zipped and uploaded.

Potential solution

Iterate through the directory that the CLI is about to try to zip and print warnings if any files are greater than 1MB.