The "situational awareness" cameras getting fitted to Abrams and things like the Ripsaws are all automotive grade or phone grade CCDs. I can buy a low rez LWIR camera that plugs into my phone for $500. It's maybe 640x480 and 30hz, but that's plenty for the purpose of not running over your infantry support. Automatic welder's hoods can cover the nucflash issue for $500 a sensor set.
Yet each module will still cost tens of thousands of dollars, because it's literally not about the sensor. It's about the fact that it's ensconced in armor grade steel, which are hand made in an artisan workshop in the United States, to pay a ballistic welder $40/hour, which an increasingly important and increasingly scarce skillset...
Also running over infantry is something that just happens, with or without cameras, because if they can't get out of the way when I reverse that's on them. Being able to look around in a tank is more for the commander's benefit than the surrounding infantrymen, but given how bad DAS is, and how glacially slow Western arms conglomerates move, I'm not sure we're close to full virtual transparency in tanks yet. Modern commercial VR headsets are still extremely novel almost a decade after their introduction.
But yes, for the primary sensor you want to get the best you can afford.
A big sensor is only useful if you don't hope to survive a hit and can engage the enemy at a tactically useful range, though.
Ukraine's weird "duct tape the camera to the back of the tank" wouldn't survive a light mortar near miss or small arms hit, let alone a major barrage in a dug in position. It's very clever for what it is but it's not very robust nor is it suitable for a military that has the luxury of peacetime to make procurement decisions. They probably don't need super FLIRs more than they just need any FLIRs at the sub-1 km distances they fight, though.
Almost all sensors on a main battle tank are bulletproof, or so redundant as to be essentially bulletproof, and the majority can't be stripped by artillery splinter. Things like atmospheric sensors and radio antennae come to mind as most vulnerable with no real easy solution for them. A FLIR's window is typically ballistic rated ceramics that can stop small arms fire, although the optic might not be able to see outside of the cracked glass very well, and the backup gunnery sights are usually buried inside the mantlet.
The best option is to simply have large amounts of sensors, apertures, or ballistic plate on hand to repair as many tanks as possible. Important wars will consume tanks by the hundreds or thousands no matter how good they are, so focusing on individual protection measures is bad if it's detrimental to the mass production, and mass field repair, of armor.
If that means skimping on a sensor then that's fine. No one complained much about the Hughes TIS in Desert Storm. It's better to have some thermal capability, even if it's poor, than to have literally nothing at all. The advantages of a cooled dual-band 720p MWIR/LWIR FLIR over a simple uncooled 240i LWIR staring array are far less than the staring array over parachute flares and image intensification.