Fix access token dictionary copy

[WMKhant]

Original dictionary is updated if copy() method is not used to copy access_token dictionary

[nullfrit]

The issue of strange behaviour of not being able to upload small file after a large file using the python sdk has been addressed with this fix.

Tested four scenarios, where small files are uploaded using upload_file and large files using upload_file_by_chunks.

• uploaded large -> large files (pass)
• uploaded large -> small files (pass)
• uploaded small -> large files (pass)
• uploaded small -> small files (pass)

Extra observations
upload_file_by_chunks() was tested having chunk size assigned from values between 10kb to 100mb (also tested 1gb chunk size) on uploading a 100mb file.

• Error message

Error splitting and uploading file with file_id ## and chunkId 1
Request experienced an internal error, please contact your system administrator referencing support id.

Occurred when chunk size was set from 30mb upwards.
Chunk sizes between 10-25mb appear to work without issue.
Setting chunk sizes too small relative to the original file size (i.e. 10kb) reduces upload speed performance, increasing upload times and likelihood of upload failure.