Sleep time not connected to usage constraints

[dannellyz]

In the job results processor the sleep on longer pdfs creates large delays. Can this instead be inferred from usage limits rather than an arbitrary setting of 5?

amazon-textract-serverless-large-scale-document-processing/textract-pipeline/lambda/jobresultprocessor/lambda_function.py

Line 28 in 2d8b97c

time.sleep(5)