* changed the download location of libdvdread to download.videolan.org because www.dtek.chalmers.se keeps refusing any connections

* removed the linkage to it in all localisations since it is a global entity now

* vlc.scriptTerminology: updated to the plist-xml-format and removed the 'localisable'-flag for the terminology in Xcode since we don't localise it and Apple won't support any other AS-dialects than English in the future --> moved out of English.lproj
* vlc.pbproj: removed localisable-flag for vlc.scriptTerminology

mux/asf.c new --sout-asf-packet-size  switch.  It defaults to 4096, current VLC design, but note that MS' documentation (http://msdn.microsoft.com/library/default.asp?url=/library/en-us/wmform95/htm/iwmpacketsizesetmaxpacketsize.asp) has: "By default, the maximum packet size is 1400 bytes (chosen because it is below the 1500-byte Ethernet maximum transition unit (MTU) plus the generic routing encapsulation (GRE) tunneling header size)."

fr.po: small translation fix

 * added some comments
 * updated module description
 * try to fix timestamping

       with the vlc-0.8.1.svn20050314 debian package)

* updated the libcaca-download-location and updated the used libdvbpsi-version to a snapshot created today

 * es_out: added ES_OUT_SET_GROUP_META to add on the fly meta for a program
 (it's not yet really clean).
 * demux/ts: parse SI tables of DVB (channel names + events), it needs last
 libdvbpsi svn (set TS_USE_DVB_SI to 1 in modules/demux/ts.c to use it).

 access2_Control/demux2_Control, use the right ACCESS_XXX/DEMUX_XXX/)
          - do not create sout when uri start by 'vlc:'. It is needed to use
 the special vlc access with sout (ex: vlc in --sout out vlc:quit).

NOTE: the slider is actually another ActiveX control but it could be replaced by a javascript equivalent if one wants a better looking slider.

  ES packets between different inputs.

  the ES ID as TS PID (contrary of --ts-es-id-pid).

(patch courtesy of Nico Sabbi)

 (Seek is slow, I don't support index yet).

reintroduce nearbyint, but use the 1001 multiplier.  adding 0.5 wasn't transforming 29.97 fps properly, and the earlier nearbyint was a hack for mpeg1/2.  the preceding mods were all partially right, partially wrong -- needed to combine the two change concepts. 

* modules/stream_out/transcode.c: reverted previous commit which didn't belong here and was also completely wrong.

* Patch to reenable sap-timeout functionality (Bug 2007) By Dermot McGahon, dermot at lincor dot com
  Modified by me.
* also restored some indentation of this file.

Thanks for the ftime-fix gibalou :)