Formally, you should indeed have a matched antenna when operating a radio transmitter, and antenna matching is indeed frequency dependent.
That said, (permissible unlicensed) LoRa RF power levels are relatively low, and the potentially lasting issue with mismatched antennas is from the reflection of transmit power back into the transmitter. The lower the power level, the less the issue. The multitech linecards already tend to run fairly hot, as they use a linear voltage regulator to drop a fairly large difference.
But even aside from the damage concerns, you'll get horrible performance without an antenna.
Fortunately, an antenna need not be complicated. It would be a relatively simple thing to design a vertical dipole for the center frequency of the subband you are using. Take a look at ham radio sites for specifics - you basically want something similar to what would be used on the 33cm ham band, only a little longer as you are on the European LoRa frequencies which are lower in frequency.
Of course, to make such an antenna you'll need an RF connector that matches the device, and a bit of coax (you don't actually need a balun, especially for a vertical dipole). It's possible you might be able to salvage this from some other type of UHF/microwave radio antenna, but do note that wifi typically uses "reverse" SMA or TNC connectors, in contrast to traditional gear which uses the traditional form. A cheap mag-mount antenna intended for RTL-SDR use could even be decent, if placed on a cookie baking pan and with the whip length adjusted per calculations.
If using an antenna in which you have limited confidence, you might see if you could configure the gateway to operate the radio at a low power level only.
An additional (and especially "polite") option if you are doing only "across the bench" testing would be to use a connectorized 50 ohm dummy load in place of an antenna. At the frequencies involved it's critical that this be a non-inductive resistor, so preferably something sold as a dummy load for frequencies of this range or higher. Power levels (especially time averaged) will be low, so it won't need to be a large device. You could even use a high ratio attenuator - as the unterminated impedance match would be only slightly off. Of course with a dummy load, your actual radiated power and pickup will be due entirely to leakage in the shielding, so range will be very short!
