Limit input length to net_data_url_fuzzer.

An earlier CL of mine made each run parse the input string twice, which
slowed things down enough to make libfuzzer (which already runs each
input through the fuzzer twice) sad.

Bug: 1019151
Change-Id: Iee963f40c9bf30a24a77961be4cef8ca64b3c20f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1903964Reviewed-by: Eric Roman <eroman@chromium.org>
Commit-Queue: Matt Menke <mmenke@chromium.org>
Cr-Commit-Position: refs/heads/master@{#713656}

net/BUILD.gn

@@ -6211,6 +6211,9 @@ fuzzer_test("net_data_url_fuzzer") {
     "//net",
   ]
   dict = "data/fuzzer_dictionaries/net_data_url_fuzzer.dict"
+
+  # IsTokenChar() and ToLowerASCII() are surprisingly slow in instrumented builds.
+  libfuzzer_options = [ "max_len=100000" ]
 }
 
 fuzzer_test("net_mime_sniffer_fuzzer") {