Switch to Reading preparing header text and processing in one call.

Currently the system spends a fair chunk of time processing input line by line.  The code path is much simplified if instead we read the maximum number of lines that we would process under the old code and process that all at once. 

This will also improve memory footprint since the full text matcher will not have to keep a copy of all the text it has seen so far, but rather just process the block of text that is passed in.