Sure you can easily write decent ftp daemon using php/tcl/... high level programming languages. However such daemon would be very inefficient. Lua is somewhat different; it's special case programming language designed to run things really fast. And due to its' outstanding performance, it's being used in many commerical applications that require high performance. eg. games and embedded devices.
It's in everyone's intrest that I spend time on optimizing 1% of code that is in use 99% of time, rather than optimizing 99% of code that is in use 1% of time. This is where ihmo others go wrong. They implement everything that users require as configuration options. Which at first it may seem like reasonable approach, but after a while you have dozens of options that many people don't even know/care about. If you had designed daemon to be easily extended - which does actually require much more time than adding those built-in options - you could have provided those same features in small scripts/modules. The only difference it makes to end-user, is that he needs to copy a file to certain location instead of modifying configuration file.
Btw, isn't it rather pointless to have this debate before you have actually evaluated the new daemon? You're welcome to say that performance sucks and that I made poor choices, if you really think that once you've tested it
