The most important advantage of setting the precision level is to be able to answer the demands of some professionals that are not satisfied with the only indication relative to precision of the 1/1,000,000 of a degree. It is indeed possible to attain precision levels well beyond the generally acceptable one millimeter.
The increased precision is generally obtained by a limitation of the scope of any map (see Example 1 above). A limitation to 2,000kms in the north-south and east-west directions will insure the 1 millimeter precision. Note that the east-west limitation is not really a constraint in "segmented" projections (UTM width is 6 degrees) because the metric coordinates are never negative and do not exceed 1,000 km.
Another advantage of controlled bound setting is the guarantee to work with a unique "basic grid"; If several projected maps from different origins and/or with different projections are piled/tiled in the same mapper, their basic grids can be in perfect alignment if similar and compatible bounds are chosen for each map. Several topological problems are avoided; for example, tracing in one layer using another as reference will give the new nodes exactly the same "internal" coordinates. - copying from one to the other will avoid any distortions, - moving nodes to have them coincide in different layers become possible whereas they may be preventing from taking the required position by non-coincidental grids.
The rules to follow are simple: for a given project, choose a precision level, translate it in a given value that leads preferably to "round" measures of precision (1 mm, 0.5 cm) and should be used for width and height, then apply the appropriate bounds to all the maps.