HTTP metrics output broken

[ztreinhart]

When running the following VHS command with the testdata/echo.bash script generating load:

./vhs --input tcp|http --address 0.0.0.0:1111 --capture-response --output json|stdout --debug --prometheus-address 127.0.0.1:8888

I expect to see VHS http specific output at the metrics endpoint at 127.0.0.1:8888/metrics. Something like this:

# HELP vhs_http_duration_seconds Duration of http exchanges seen by VHS.
# TYPE vhs_http_duration_seconds summary
vhs_http_duration_seconds{code="200",method="GET",path="/",quantile="0.5"} 0.0003203
vhs_http_duration_seconds{code="200",method="GET",path="/",quantile="0.75"} 0.0005124
vhs_http_duration_seconds{code="200",method="GET",path="/",quantile="0.9"} 0.0006848
vhs_http_duration_seconds{code="200",method="GET",path="/",quantile="0.95"} 0.0007811
vhs_http_duration_seconds{code="200",method="GET",path="/",quantile="0.99"} 0.0009974
vhs_http_duration_seconds{code="200",method="GET",path="/",quantile="0.999"} 0.0011647
vhs_http_duration_seconds{code="200",method="GET",path="/",quantile="0.9999"} 0.0011647
vhs_http_duration_seconds_sum{code="200",method="GET",path="/"} 0.1551513
vhs_http_duration_seconds_count{code="200",method="GET",path="/"} 412
# HELP vhs_http_requests_total Total count of http requests captured by VHS.
# TYPE vhs_http_requests_total counter
vhs_http_requests_total{code="200",method="GET",path="/"} 412

However, when using a VHS binary built from the current main branch, there is no VHS HTTP data in the metrics output, only standard output from the Prometheus library.