David, what you propose would work. And, yes, it is a workaround, but it is at least a reasonably effective one.
What I actually DO is one of two things....
1) Just visually line up (the misalignment is invisible) and anchor the part.
2) Align the part using one of the sides of the part plus an offset of the side-to-hole amount.
If I understand you, the second is equivalent to your approach. And, all three are, indeed, "workarounds". Two have the difficulty that they only align lengthwise, and do not actually constrain the part in its position. That is not always sufficient. Anchoring has the difficulty of turning constraints red in the explorer, which screams that the sky is falling, even though it is not.
I just want a constraint available that acknowledges the existence of tolerances, so that I can constrain a part that fits fine and is within-tolerance.
Lew:
I do not want it only as a check, but f I understand you, your idea is a useful tool.
I do not quite see the "using constraints improperly". If I understand the "best practices" folks, ANY use of an actual part-to-part constraint is "improper".
The ONLY acceptable use of constraints to some folks is to constrain to a complex network of reference planes, lines and points, upon which the parts are to be "hung" in a manner similar to reconstructing a crashed aircraft for examination (and it is a valid idea FOR SOME USES). Use to constrain to an actual part feature is regarded as beyond the pale, the mark of a clueless neophyte.
Not all usages require that approach. Some do.
The idea of the tolerant constraint brings real tolerances into the model, it is not some sort of cheat to make things appear to work. If anything, the workarounds cited above are the real "cheats", since they will "work" even in the case of serious errors.
Folks have commented to me off-line that they do not like Alibre for steel structure work for basically the reason I have mentioned. And, in some cases, that the tolerant constraint would make things easier to deal with, while not compromising the design.
What is "wrong" for your work may not be (I assert IS NOT) wrong for other work. Tools can be used or abused.
I personally assert that adding an axial plane to constrain to, etc (the aforementioned workarounds), is an abuse, one likely to result in parts that fail to fit.
I assert that, in contradistinction to the usage expressed in the last sentence, use of a tolerant constraint would be a very proper procedure,
I further assert that Alibre is evaluating constraints in an improper manner, and that it should restrict evaluation to the 6 visible decimal places. That applies a de-facto tolerance, but it is one I defy anyone to measure or observe reliably outside the calibration lab, in the real world. Even YOUR world.