doc: improved error reporting for non-UTF8 docs

Improve error reporting to include the filename being processed when a
problem occurs during a pre-generation phase when boards and samples
files are temporarily copied (by doc/scripts/extract_content.py) into
the documentation area for processing.

Two recent problems were noticed:

  Some new files were using window-1252 encoding and included windows
  printer quote marks and hyphens, for example 0x92 in window-1252
  encoding is Unicode 0x2019 for 'RIGHT SINGLE QUOTATION MARK'.

  An image file reference by a reST file was missing

Both of these threw an exception reporting the error, but did not
include any information about the file currently being processed, making
it hard to fix the problem (e.g., change the Windows right quote
character to an ASCII ').

Signed-off-by: David B. Kinder <david.b.kinder@intel.com>