the dependency at this point is implemented as a RPM dependency on the scalix-server package, so there is technically no way to make this optional or conditional on a switch. You can use some --force option on the rpm command to install this. If you want to do this in the context of scalix-installer, you can also mod that as it is open source and python-based.
Now, the bigger question is - why would you want to do that?
I will not go into any discussion about sendmail vs. postfix, this is really very much a matter of personal preference. Even the distros disagree, and while SuSE and debian-based distros usually have postfix setup as a default, everything that's Red Hat-based (RHEL, Fedora, CentOS) comes with sendmail enabled as a standard.
The other arguments don't seem to be too relevant. I consider sendmail.mc my config file, everything else is a parsed/generated config that I don't care about, so it's a single place where I do my changes. Actually in Postfix these days you are not supposed to edit config files directly but use the command line tool to do the validation. The remaining arguments don't seem to fly either - for me the only config I need to do on the sendmail side in 99% of all cases is the SmartHost, which is a single-line config and very straightforward. Authentication is not necessary as Scalix uses sendmail only as it's MTA for OUTGOING email - inbound stuff and message submission is handled by the Scalix SMTP Relay, which is preconfigured for authentication and else. Since 11.4, even AntiSpam/AntiVirus setups don't require sendmail mods anymore as we have implemented the MILTER interface right on the SMTP relay.
For this reason, I fail to see any specific advantages for Postfix in a Scalix environment. On the downside, we could not simply drop support for sendmail if we chose to support Postfix - too many installations out there that know how to monitor, operate and configure it. And supporting both is simply a matter of resources on our side - not only would we need to implement the integration, but we'd also have to maintain it, and, and that's the biggest obstancle, run all our server-side testcases on every platform on both MTAs. That would actually double our testing matrix, which is already horrendously large and it would make us spend more money on something that's not highly useful instead of working on relevant new features. Or it would make new Scalix releases and patches take longer to get out of the building, again a huge disadvantage in my book.
So ... I'm yet to be convinced on this one!
Florian.