fix BigInt ndigits test

See #16766. `ndigits(big(0),1)` doesn't raise an error, but we were
sometimes (randomly, rarely) testing that it does. This change makes
the test reliable. The issue of how this case should actually
behave is still open.

test/bigint.jl

@@ -278,7 +278,11 @@ ndigits_mismatch(n) = ndigits(n) != ndigits(BigInt(n))
 ndigits(rand(big(-999:999)), rand(63:typemax(Int)))
 ndigits(rand(big(-999:999)), big(2)^rand(2:999))
 
-@test_throws DomainError ndigits(rand(big(-999:999)), rand(typemin(Int):1))
+for i in big([-20:-1;1:20])
+    for b in -10:1
+        @test_throws DomainError ndigits(i, b)
+    end
+end
 
 # conversion from float
 @test BigInt(2.0) == BigInt(2.0f0) == BigInt(big(2.0)) == 2