Fix to apply conversion rules to data within data frames (if raw is not enabled)

[nickpelling]

A co-worker (Jiri) was trying to use the API's export() function to convert MDF data to MAT files, but noticed that channels inside a dataframe were always being exported as raw data (i.e. the channel's conversion wasn't being carried out). He asked me if I could tweak export() to do the conversion.

Here's the code I added (I put it just before the mdict.update(sigs) call in mdf.py):

                            if raw == False:
                                siglist = list(sigs.items());
                                for k in range(len(siglist)):
                                    conversion = self.groups[i].channels[2+k].conversion
                                    if conversion:
                                        cur_tuple = siglist[k]
                                        converted_data = conversion.convert(
                                            cur_tuple[1],
                                            ignore_value2text_conversions=ignore_value2text_conversions
                                        )
                                        sigs.update({cur_tuple[0]: converted_data})

Now, I'm pretty sure that this isn't perfect (the "2 + k" is somewhat clunky), but it does seem to do the job for the MDF data I tried it with: and if you invoke it with raw(True), you get the existing behaviour. Should this (or something not too dissimilar to it) be delivered to the repo?

Note: I believe that this is also true of the CSV exporter, so something similar should probably be deliverered into that export() path.