powershell : Improve the pipe reading

Previously in MODULES-8358 it was found there was a slight bug in the pipe reader in that it
wasn't reading all of the data.

A workaround was put in place as part of comit https://github.com/puppetlabs/puppetlabs-powershell/pull/247/commits/b8db256080b7a7f138b1e2ab13c5e4c25375fe78 however ethan raised some valid concerns over the workaround

This ticket will track the work to look at using a different pipe reading method which isn't subject to same issues, namely:

• Just because I chose a chunk of 8K doesn't mean that that is small enough.
• Concatenation can be an inefficient method of memory allocation. There should be better ways

Conversation reference:
https://github.com/puppetlabs/puppetlabs-powershell/pull/247#discussion_r261334543

Commit
https://github.com/puppetlabs/puppetlabs-powershell/pull/247/commits/b8db256080b7a7f138b1e2ab13c5e4c25375fe78