I had similar problems with my CRS305 and decided to do some power draw testing.
I bought 4x S+RJ10 and 4x 80m 10GBase-T modules (
Cisco SFP-10G-T-80 Compatible 10GBASE-T SFP+ Copper RJ-45 80m Transceiver Module - FS Germany)
When the CRS305 is idling, no cables connected, it draw 3,312W (0.138A * 24V). That is much lower than the max draw of 8W stated by Mikrotik.
I then connected 1 Gb/s RJ45 to the MGMT port and added 4x S+RJ10 (one by one, noticing the increased power draw).
When the switch was pushing 10 Gb/s on all four ports using 4x S+RJ10 the max noted power draw was 19,1W.
I could see that the power draw increased some when the switch and modules got hotter (from 18W to 19,1W).
All ports worked as expected, but I was only running the tests for ~30 min, so I can't say anything about long term performance.
Without a fan the switched reached 55 C (unit temp according to web gui) and the modules reached 94 C.
When I added a 120mm fan on top of the unit the temp went down to 32 C for the unit and 58 C for the SFP+ modules.
My conclusion is that you can actually use the CRS305 with four S+RJ10 modules if you are cooling it. The power supply (24V / 1,2A) should be enough.
I then went on doing the same tests with the Broadcom based (I think) 10GBase-T SFP+ modules from FS.com (I used the FS.com modules, but you can problably buy any "80m" RJ45 SFP+ module and get the same result).
When using four SFP+ modules at 10 Gbit/s and a 1 Gbit/s link in the MGMT port the switch reached a maximum power draw of 13,1 W.
The unit temp went up to 55 C during my 30 min test. The SFP+ modules does not report it's temperature, but I could tell they were much cooler than the Mikrotik modules when not using a fan. I'm guessing ~65 C for the modules.
This is of course a much better setup then the power hungry Mikrotik modules (not considering the higher price, that is).
Based on changes in power draw when adding/removing/using SFP modules I would approximate the power draw like this:
CRS305: 4W with link on MGMT port
S+RJ10: 3,5W
FS 10G-T-80: 2,3W
One interesting result is that the unit could actually exceed it's rated max power draw by more than a watt.
Another result is that the performance (throughput using iperf3) was not affected by heat (90+ Celsius SFP units).
I hope these tests can provide some help for you when deciding which hardware to choose.