Stream query file contents to kusto

Instead of reading the specified --file into memory as a string to assign it to the kusto HTTP request body, stream it directly from the file to the request body in order to reduce memory use. Might be better to only do this for files above a certain size (e.g. 4KB?) but also might be easier to just do it all the time. The goal here is to reduce memory use and improve performance of the CLI when executing a large query.

This might impact the Kusto query parsing and syntax validation we do before sending the request though so that needs some thought, e.g. can we do the checking in chunks, should we support disabling the checking or opting into streaming mode (which would disable syntax checking), or support disabling the syntax checking, etc.

Trace the CLI's --file handling through construction of the Kusto HTTP request, starting where the file is read into a string. Determine how streaming affects Kusto query parsing and syntax validation, then define behavior for large files and validation before measuring memory use with a large query. Done means file contents stream to the request without the current full in-memory string and the validation behavior is explicit.