As I understand it, the power that a steam engine produces is a function of rpm, steam pressure, and the timing of the valve events.
The force generated on the piston rod is a function of the steam pressure acting on the top of the piston, and the area of the piston (pressure/area).
The critical valve timing events are admission, cutoff, release, and compression.
For small model engines, I have seen all sorts of bad valve timing work sufficiently for the engine to run well without load.
For locomotives, getting the valve timing correct may mean the difference between pulling the passengers and cars up a grade, or not having sufficient power for that work.
Also important is equalizing the valve timing events on both the instroke and the outstroke.
Some odd situations like large (full size) vertical engines use asymmetrical valve events to take into consideration gravitational forces.
For engines with Stephenson's links, the valve timing needs to be checked to make sure that the events are the same in forward and reverse.
I don't make the clearance of my model steam engines too tight, since they are not generally designed to pull a load (my future designs will pull a load, but even those will not be too tight).
The problem with clearances that are too tight is that the bearing surfaces will wear, especially at the connecting rod bearings, and if the take-up adjustment is not done correctly, the connecting rod will get shorter or longer as the wear is taken up, and this can cause the piston to crash into the upper or lower head.
The ideal connecting rod bearing arrangement is one that results in zero change in length as the bearing wear is taken up.
Efficiency may come into play for a locomotive, but for a model engine running without load, I don't think you will notice any efficiency changes with clearance changes. There may be efficiency changes due to clearance at this scale, but they would probably be difficult to measure.
Launch engines are operated under load, and so the launch folks tend to pay much more attention to engine design and sequence of valve events than model engine builders. Efficiency could be important for a launch engine, but I have seen people opt for a two cylinder non-compound over a compound design, just because the don't care about the fuel efficiency of a small launch, and they want the simplicity of a non-compound engine.
Others insist on a compound arrangement since they may be limited in how much fuel they can carry.
As Richard mentions above, excess clearance is just wasted space that has to be filled and emptied with each stroke, and the extra steam required to fill this space does not provide any additional pressure on the piston, and does not provide any more power to the engine as far as I understand it.
For full sized engine designs I have seen, the piston does not generally intrude into the passage space at TDC and BDC, else piston slap can occur.
The interior of the heads can be adjusted during the machining phase to protrude more down into the cylinder (assuming you are making cylinder covers from scratch), but the protrusion space on the cylinder head at the passage area (for tight clearances) is generally milled to a round shape, to provide a ramp entry of steam into the cylinder, and to prevent the protrusion from blocking the passage.
For steam engines that do not use a D-valve, sometimes a relief valve is fitted to the upper and lower cylinder heads.
A D-valve will lift off of its seat and relieve a hydraulic situation where you may get an unintentional charge of water into the cylinder.
For a non-D-valve, you can bend/break things if you get condensate in the cylinder, especially if you are running very little clearance in the engine, since some valve types will not automatically relieve a hydraulic lock situation.
As a side note, the piston rings on large steam engines generally overruns the end of the bore slightly to prevent ridge buildup.
For maximum efficiency, an early cutoff is used.
For maximum power, a late cutoff is used, such as a locomotive trying to start on a grade from a standstill with a heavy load.
The Stephenson link allows the cutoff to be varied dynamically to suit the speed of the engine, and greatly improve efficiency.
The Stephenson link also has the critical advantage of advancing the timing (advancing the admission of steam), which is similar to what an automotive engine ignition does.
.
