tags: don't scan files twice.

* tags.tl (collect-tags): Snarf the file as a string from
which we can make a string byte input stream suitable for
parsing. Having that string, we can handle hash bang by
detecting it and commenting it out with a semicolon. We obtain
the lines by splitting the string. When calling read, we
pass the path as the name argument, so error messages still
use the correct file name.

0 files changed, 0 insertions, 0 deletions