Possible bug in RegionDistance with RegionBoundary?

Code as simple as it is:

targetRegion = RegionUnion[Disk[{0, 0}], Disk[{1.8, 0}]];
edgeDistFn = RegionDistance[RegionBoundary@targetRegion]
pts = RandomPoint[targetRegion, {2, 2}];
pts // edgeDistFn

This should give distances of all points to the boundary, but works on neither 12.1.1 nor 12.2, with output like:

{{RegionDistance[
   RegionBoundary[
    BooleanRegion[#1 || #2 &, {Disk[{0, 0}], 
      Disk[{1.8, 0}]}]], {-0.4243129646, 0.5777295089}], 
  RegionDistance[
   RegionBoundary[
    BooleanRegion[#1 || #2 &, {Disk[{0, 0}], 
      Disk[{1.8, 
        0}]}]], {1.227762454, -0.3983038364}]}, {RegionDistance[
   RegionBoundary[
    BooleanRegion[#1 || #2 &, {Disk[{0, 0}], 
      Disk[{1.8, 0}]}]], {2.160377018, 0.6768344764}], 
  RegionDistance[
   RegionBoundary[
    BooleanRegion[#1 || #2 &, {Disk[{0, 0}], 
      Disk[{1.8, 0}]}]], {1.547838415, -0.00215747549}]}}

Is this a bug or an individual issue?

Despite @cvgmt’s answer (a makeshift, inaccurate, and numerical solution), the reason why we need to discretize the region (in many cases, especially those where $mathtt{Disk}$s, $mathtt{Cone}$s, etc., are involved in $mathtt{RegionUnion}$, $mathtt{RegionIntersection}$, etc.) or most of the functions for region measurement will fail to work is still unclear.