Some tag encoders like to use non-standard padding (see http://id3.org/Compliance%20Issues) which was causing "unexpected EOF" errors. Checking that each tag has a valid name (as per the spec) seems to be the only good way around this.
I hacked readTextWithDescrFrame() and readTFrame() to generalize them
to other case (presence or not of a language and whether or not the
data part is encoded (URL are not)
This patch depend of 3b7a3d5485 since I found the
bug while developping this patch
Sometimes, we can have this configuration :
description 00 00 00 data (with delim = [00 00])
with splitN, we make the separation at the first 00 00 :
[description, 00 data]
But in fact, we must make the separation one byte later
[description 00, data]
