[DOC] document using `grofer export` format

[MadhavJivrajani]

• The format in which metrics are exported by grofer export resembles the ndjson format.
  • This is done for ease of appending json objects to a file without having to buffer large json strings in memory.
  • It also makes it easier to read the file in efficient ways such as by maybe lazy loading using generators in python.
• It'd be helpful to have this mentioned in greater detail in the README and also maybe have a few examples that make use of these metrics that are exported as ndjson showing how such files can be worked with.

Getting metrics

grofer export -p <valid PID> -f export.json

[Samyak2]

I would like to point out that ndjson isn't technically valid JSON since a JSON file contains just one object and not multiple. As per the ndjson spec, the file name extension should be .ndjson and not .json.

This part in the export help page should be fixed to say ndjson instead of json.

  -t, --type string       specify the output format of the profiling result (json by default) (default "json")

Adding this here since this is also an issue in the documentation.