Depends where they are on the bike.
I copper slip the bolts holding my arm rests on as they are likely to get salty sweaty. Other than those I go by the manufacturer of the parts recommendations which may be for grease, locktite, etc.
I regularly maintain my bike and use a torque wrench though.
I locktite my viewspeed skewers as that’s what’s Recommended.

The gold standard would be that the design organization would

1) perform torque tension tests with the the intended design (materials, lubricant, where lubricated, etc)
2) based on the testing from 1) determine the mean preload and the inherent variability
3) on whatever engineering documentation (drawings, process documents) specify installation requirements (torque, where/what lubricant, materials in the clamp, etc)

But bikes, particularly once you deviate from OEM hardware, don't adhere to that gold standard. What you found I your workshop is accurate. Cleanliness, lubricated vs not, where lubricated, etc all affect the torque preload relation. Clean (i.e., stripped of any wet/dry lubes or surface treatments) and dry bolts will have a higher nut factor than lubed, meaning for a given torque dry has less preload. Along with that is that dry will have more scatter about the mean. Or say preload=(1 +/- scatter ) × torque / (diameter x mean_nut_factor). That scatter will be higher for dry. Things like loctite or other locking compounds can really drive up the scatter. Even things like reusing the fastener can affect performance. As you install and remove the fastener the sliding motion in the threads generates wear particles which will increase friction in the threads. In turn, this increases the nut factor (generating less preload for a given torque) and increases scatter.

On a bike you have to think of what functions the bolts serve. In general you will you bolts in strength, preload, or alignment critical applications. Think of all the fasteners on your bike. Most are preload critical in slip - you don't want parts slipping. You address slipping increasing preload and/or increasing friction at the slip surface. Intuitively you may think about just cranking up the preload, but then think about the materials in the stack. What if you have a carbon frame and seat post versus metallic frame and post? Steel vs aluminum? Clearly a one size fits all torque doesn't work. In strength critical applications, if you were to overload the joint the joint will likely separate prior to the fastener breaking. In such a case you start plasticly yielding the fastener, dropping preload. In essence preload doesn't matter so much.

I don't own a torque wrench. I have never had an issue aside from joint slipping on composite parts. Knowing what I know I want to dirty up the slip plane as much as possible to increase friction. I also want a higher friction in the threads. Yes I have more scatter, but my mean preload will be lower. This allows me to slowly creep up on a clamp force. I go by feel (not for everyone). Never ruptured any composite parts in 20 years. On many other parts of a bike, they are there to merely attach items and the service loads are well below the capacity of the fastener. The issues there can be self loosening from vibration. In such case higher preload to prolong the time it takes to self loosen or a locking compound does the trick. By the way, locking compounds or locking features don't necessarily prevent preload loss. I could point you to a great thesis surveying various locking features and their performance in self loosening. I would also recommend looking at NASA-STD-5020 which you can find online to get a feel of someone the discussion above. I am one of the authors and routinely work fastener related issues in the aerospace sector.

I generally use a touch of grease. Not so much to worry about torque, but various types of corrosion, particularly if aluminum is involved. But also steel alloys.

I have a torque wrench, but often just go by feel as well.

Aside from all of the technical torque/preload/bolt physics, which really is very cool, there is also a practical standpoint.

The only time you're really rebuilding a bike is after travel, and the parts you're rebuilding are: pedals, stem/steerer, seatpost.

The pedals aren't coming out unless you pedal backwards. Grease is just soo they don't get stuck. The threads are so coarse that basically isn't going to happen.

The stem/steerer has two bolts of the most critical bolts on the entire bike. Too tight and you can damage the carbon steerer, too loose has obvious problems. All that said, these bolts only come out a turn or two, so the internal grease remains consistent in all but the worst of conditions. Just tighten to the stated torque on the stem.

Ditto for seatpost.

Good answer. I would just note that the tool used (and who is using the tool) will also have a big impact and that dry threads or high friction can go wrong with some material combinations resulting in galling/cold welding. Plenty of stuck titanium bolt stories. In that case, in combination with aluminum, galvanic corrosion can also be an issue.

Bolt condition and prep definitely makes a difference.

What's interesting is that "feel" can be counterintuitive. An old dry bolt with a bit of thread rust will often slip very smooth through a nut, and then feel like it's making a very firm and snug action as it engages. It "feels good" to a lot of people. But that smooth slipping through the nut wasn't because the bolt had low friction, it was because the corrosion loosened its tolerances over time. And the snug feel wasn't the bolt rapidly reaching high axial force, it was the rust causing the thread friction to create a huge amount of resistance with even very slight pressure on those surfaces.
This can, for instance, lead to confusing seatpost slip on old bikes. The mechanic becomes worried that they'll crush the post because they're torqueing the bolt so hard to combat slippage, when the post is barely being clamped at all.