Add the ability to read file headers from custom record extractors

[yruslan]

Background

Sometimes information defined in file headers is needed for processing files themselves.

Currently, custom record extractors receive input streams of data from the index location, which can be in the middle of a file.

Feature

Add the ability to read file headers from custom record extractors.

Proposed Solution

Add another stream to the extractor context. The second stream should always point to the beginning of the file.