adds loose_unicode option which replaces badly formed unicode (escaped non-characters and restricted codepoints, as well as partial surrogates) with u+fffd instead of throwing an error

test/cases/escaped_noncharacter_ext_replaced.json

@@ -0,0 +1 @@
+"\ud83f\udfff"

test/cases/escaped_noncharacter_ext_replaced.test

@@ -0,0 +1,4 @@
+{name, "escaped noncharacter (extended)"}.
+{jsx, [{string, <<16#fffd/utf8, 16#fffd/utf8>>}, end_json]}.
+{json, "escaped_noncharacter_ext.json"}.
+{jsx_flags, [loose_unicode]}.

test/cases/escaped_noncharacter_replaced.json

@@ -0,0 +1 @@
+"\uffff"

test/cases/escaped_noncharacter_replaced.test

@@ -0,0 +1,4 @@
+{name, "escaped noncharacter replacement"}.
+{jsx, [{string,<<16#fffd/utf8>>},end_json]}.
+{json, "escaped_noncharacter_replaced.json"}.
+{jsx_flags, [loose_unicode]}.

test/cases/escaped_nullbyte_replaced.json

@@ -0,0 +1 @@
+"\u0000"

test/cases/escaped_nullbyte_replaced.test

@@ -0,0 +1,4 @@
+{name, "escaped nullbyte replaced"}.
+{jsx, [{string,<<16#fffd/utf8>>},end_json]}.
+{json, "escaped_nullbyte_replaced.json"}.
+{jsx_flags, [loose_unicode]}.

test/cases/noncharacter.json

@@ -0,0 +1 @@
+"﷐"

test/cases/noncharacter.test

@@ -0,0 +1,3 @@
+{name, "noncharacter"}.
+{jsx, {error, badjson}}.
+{json, "noncharacter.json"}.

test/cases/unpaired_surrogate.json

@@ -0,0 +1 @@
+["\ud801blah"]

test/cases/unpaired_surrogate.test

@@ -0,0 +1,3 @@
+{name, "unpaired_surrogate"}.
+{jsx, {error, badjson}}.
+{json, "unpaired_surrogate.json"}.

test/cases/unpaired_surrogate_replaced.json

@@ -0,0 +1 @@
+["\ud801blah"]

test/cases/unpaired_surrogate_replaced.test

@@ -0,0 +1,4 @@
+{name, "unpaired surrogate replaced"}.
+{jsx, [start_array,{string,<<16#fffd/utf8, "blah">>},end_array,end_json]}.
+{json, "unpaired_surrogate_replaced.json"}.
+{jsx_flags, [loose_unicode]}.