Indeed... I'm sorry, but I can't afford spending time building 8 binaries (2 viewer branches, 2 OSes, 32 and 64 bits for each) each week, not to mention the 400Mb worth of added data they would take up against my quota on my ISP's web server (which I use for various sites, and not just the Cool VL Viewer's), and maintaining 4 different build systems (Linux and Windows, 32 and 64 bits).
So a choice must be made, and since all modern OSes (*) and computers are now 64 bits, the move is unavoidable.
Also "long" is inappropriate an adjective: the move will happen in the short
term (Linux first, then Windows, and hopefully MacOS-X later).
However, the sources (and pre-built libraries) will still allow to build 32 bits viewers on 32 bits OSes, but you will have to do it yourself (it's dead easy for Linux binaries).
(*) And modern Linux distros seem to all be crippling their 32/64 bits "mulitlib" support (which used to work in older distros); for PCLinuxOS 2016 which I migrated to, I had to recompile and repackage myself several libraries so that both the 32 and 64 bits versions could be installed and live along each others: the fault of broken package naming conventions (e.g. naming the 64 bits xcb libraries package "libxcb" instead of "lib64xcb", causing a name conflict with the 32 bits "libxcb" package) and broken directory tree conventions (such as putting 64 bits libraries in /usr/lib instead of inside /usr/lib64 !).