Always pass default precision to BigDecimal when parsing Float in XmlMini

https://github.com/ruby/bigdecimal/blob/cb2458bde33bf90a8364b58d53e8948a7ba555ea/ext/bigdecimal/bigdecimal.c#L2747-L2749

Author: zzak

activesupport/lib/active_support/xml_mini.rb

@@ -73,6 +73,8 @@ def content_type
         "decimal"      => Proc.new do |number|
           if String === number
             number.to_d
+          elsif Float === number
+            BigDecimal(number, 0)
           else
             BigDecimal(number)
           end