I have a program called SEvoNA (Simulated Evolution of Nucleic Acid) which
gives lots of insight into mutational processes. It generates a random or
GC-biased DNA sequence up to 1000 bp long and allows it to evolve in either a
pectinate or symetrically dichotomous distribution to give up to 32 progeny
sequences. distances between the nodes can be equal or variable. Any degree of
transition bias is allowed. Fossil taxa are allowed, or you can remove taxa to
effectively create any topology you like. Different rates of evolution are
allowed down one half of the tree. Sequences can be analysed for
phylogenetically informative sites (the frequency of homoplasies will amaze you
and humble 'solve-it-all' gene jocks !) or to produce distance matrices, output
formatted appropriate for immediate analysis by a number of popular
phylogenetic analysis packages. Its shareware, so I'm happy to let you have it.
One of my other programs, DAPSA (Dnaand Protein Sequence Alignment and
analysis) is a very user-friendly packege which might help your point 4.
Finally, Mike Gilpin, Dept of Biology, Univ. of San Diego, has a range of good
pop. genetic sim. programs, including genetic drift.