BioEdit is probably the single best tool except that it is no longer being supported. I believe there was such a tool in the Wisconsin package as well. Sequencher also provides some good tools.
The real issue is what manual editing is being used for, however. I'm not sure that the community has produced a good MSA program that duplicates the key functions of most manual MSA editing... if it has, I don't know about it. Those functions would allow locking in 'what is known' - reference alignments, boundaries of coding regions and annotated regulatory sequences, etc - so that coding regions are aligned by codon preferentially, for example, while adjacent non-coding regions are not; also it would take quality scores and reference trees, cases where reads come from the same region of the same template, for minor 'assembly' solving.
Hopefully this suite of functions would largely supplant manual editing and especially Sanger base calling.