BigFloat: fix inconsistency in allowed precision (JuliaLang#40872)

Fixes JuliaLang#40538

Author: RohitRathore1

base/mpfr.jl

@@ -815,8 +815,8 @@ Set the precision (in bits) to be used for `T` arithmetic.
     setting.
 """
 function setprecision(::Type{BigFloat}, precision::Integer)
-    if precision < 2
-        throw(DomainError(precision, "`precision` cannot be less than 2."))
+    if precision < 1
+        throw(DomainError(precision, "`precision` cannot be less than 1."))
     end
     DEFAULT_PRECISION[] = precision
     return precision

test/mpfr.jl

@@ -475,7 +475,8 @@ end
     @test precision(z) == 240
     x = BigFloat(12)
     @test precision(x) == old_precision
-    @test_throws DomainError setprecision(1)
+    @test precision(setprecision(1) do; BigFloat(23); end) == 1  # minimum-precision
+    @test_throws DomainError setprecision(0)
     @test_throws DomainError BigFloat(1, precision = 0)
     @test_throws DomainError BigFloat(big(1.1), precision = 0)
     @test_throws DomainError BigFloat(2.5, precision = -900)