Handle cut-off UTF-8 sequences when reading files

Read additional bytes from the file to complete the UTF-8 sequence so the bytes
in it don't get converted to U+FFFD replacement characters.

tests/utf8test

@@ -0,0 +1,10 @@
+#!/bin/sh
+
+. "${0%/*}/setup" "$@"
+
+if [ "`$VALGRIND $Q $JQ -nf $JQTESTDIR/utf8-truncate.jq`" != "true" ]; then
+	echo "UTF-8 byte sequences that span the jv_load_file read buffer are mangled"
+	exit 1
+fi
+
+exit 0