Hi,
I believe there is some confusion here...
Firstly, a latch implies that there is no clock input and so the design is asynchronous anyway. I don't see how the latch will infer input data from asynchronous input unless you make direct wiring...
Secondly, with RTL design methodology, you better avoid using latches altogether. I have been using edge-triggered flipflops only(for the last 10 years!!). I don't think any FPGA engineer will dare use asynchronous design at any point in their design. ASIC may be. I understand a new very fast FPGA has been announced that uses asynchronous internal architecture but appears as synchronous from the desiger's perspective.
Thirdly, the initial condition of '0' is only relevant to a simulation model and is a trivial tool-dependant issue.
Regards