(svn r14564) -Fix (r14555): one more incorrect use of lastof()

(svn r14563) -Fix [FS#2395]: in the case that an industry NewGRF, a shared TTDPatch and
OpenTTD feature with it's origin in TTDPatch to replace/add/change vehicles
(including e.g. maximum speed, graphics and introduction year), stations,
bridges, industries, town houses or any other graphics used by either
TTDPatch or OpenTTD, would tell that building an industry is okay, which
is queried using a so-called callback that allows the NewGRF author to
test all kinds of information about the neighbourhood where the industry
might get build, it would corrupt, i.e. overwrite with invalid data, the
structure that is used to build up an error message. This then might result
in trying to resolve an invalid StringID, an internal numeric representation
of all translateable strings, which would in it's turn trigger a safety
check that is added to ensure invalid StringIDs are never resolved.

(svn r14562) -Fix (r14561): 8 bits are enough. The rest stays reserved.

(svn r14561) -Feature(ette)[FS#2334]: Add result 0x0F to callbacks 0x29/0x35. (Yexo)

(svn r14560) -Fix [FS#2396](r14555): lengthof() can't be simply replaced by lastof() in some cases (part by Aali)

(svn r14559) -Fix [FS#2387]: A train could be blocked inside a depot if it was reversed just after leaving the depot.

(svn r14558) -Fix (r13957): Converting the track type of a tunnel/bridge could cause trains to get stuck.

(svn r14557) -Codechange: Remove a redundant line of code

(svn r14556) -Fix (r14533): Bounding box with height > TILE_HEIGHT could cause circular dependencies with surrounding bridges.

(svn r14555) -Codechange: replace ttd_strlcat and ttd_strlcpy with strecat and strecpy where direct conversion is possible