Link upstream issue for LQ regression on GPU-backed inputs

Reference QuantumKitHub/MatrixAlgebraKit.jl#218 from the comment on
the broken `c * u ≈ a` assertion so the marker has a paper trail when
the upstream fix lands.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: mtfishman

test/test_abstract_blocktype.jl

@@ -105,7 +105,8 @@ arrayts = (Array, JLArray)
         c, u = f(a)
         # `right_orth`, `lq_compact`, `lq_full` on JLArray-backed
         # `BlockSparseMatrix{T, AbstractMatrix{T}}` produce a `c * u` that does not
-        # reproduce `a` (regression in MatrixAlgebraKit 0.6.6).
+        # reproduce `a` (regression in MatrixAlgebraKit 0.6.6, tracked at
+        # https://github.com/QuantumKitHub/MatrixAlgebraKit.jl/issues/218).
         @test c * u ≈ a broken = (arrayt ≢ Array && f ∈ (right_orth, lq_compact, lq_full))
         if arrayt ≡ Array
             @test isisometric(u; side = :right)