Stop rewriting \A to literal A in regexes

I introduced this behavior in: #12
but thinking about it again, it seems more annoying than it is helpful.
If someone writes `\A` in their regex, they're probably writing it in
Ruby with the expectation that it's beginning-of-string. It seems less
likely that they're writing an ECMA 262 regex with the expectation that
it matches a literal `A` (because why would they write it `\A`?).

Author: davishmcclurg

lib/json_schemer/schema/base.rb

@@ -25,8 +25,6 @@ def merge(
 
       RUBY_REGEX_TYPES_TO_ECMA_262 = {
         :anchor => {
-          :bos => 'A',
-          :eos => 'z',
           :bol => '\A',
           :eol => '\z'
         },

test/json_schemer_test.rb

@@ -904,10 +904,10 @@ def test_it_handles_regex_anchors
     refute(schema.valid?("foo\nfoo\nfoo"))
 
     schema = JSONSchemer.schema({ 'pattern' => '\Afoo\z' })
-    assert(schema.valid?('Afooz'))
-    refute(schema.valid?('foo'))
-    refute(schema.valid?('Afoo'))
-    refute(schema.valid?('fooz'))
+    assert(schema.valid?('foo'))
+    refute(schema.valid?(' foo'))
+    refute(schema.valid?('foo '))
+    refute(schema.valid?("foo\nfoo\nfoo"))
   end
 
   def test_it_handles_regexp_resolver