Add an XXX note concerning an HTTP/2 test for e00019.vtc.

Repeating the test in e00019.vtc with HTTP/2 fails both here and
for standard Varnish. My understanding of HTTP/2 is too poor for
me to know whether there is a bug in a Varnish, or if I am not
applying HTTP/2 correctly. More details in the comments of the vtc.

Setting this aside for now, but we should come back, but we should
come back to it (and if necessary file a Varnish bug report).

src/tests/e00019.vtc

@@ -83,3 +83,35 @@ logexpect l1 -wait
 
 varnish v1 -expect esi_errors == 5
 varnish v1 -expect esi_warnings == 1
+
+##
+## XXX: repeating this test with HTTP/2 fails both here and for
+## e00019.vtc in standard Varnish. My understanding of HTTP/2 is too
+## poor for me to know whether there is a bug in a Varnish, or if I am
+## not applying HTTP/2 correctly.
+##
+## From the log (running the test both here and in standard Varnish),
+## it appears that the stream stops transmitting after sending 16 data
+## frames for a total of 64k-1 bytes. Those two numbers (16 & 64k)
+## look suspiciously like default settings for something. But various
+## attempts on my part to change h2_* params, or to use txsettings to
+## change the winsize, for example, did not lead to success.
+##
+## This suggests that the problem might not be caused by pesi --
+## whatever it is, e00019.vtc in standard Varnish has the same
+## problem. So I'm setting this aside for now, but we should come back
+## to it (and if necessary file a Varnish bug report).
+##
+
+## HTTP/2
+
+# varnish v1 -cliok "param.set feature +http2"
+
+# client c1 {
+# 	stream 1 {
+# 		txreq  -url bar
+# 		rxresp
+# 		expect resp.status == 200
+# 		expect resp.bodylen == 65856
+# 	} -run
+# } -run