Enforce HTTP timeouts on a per-read (rather than per-request) basis #5222

Job	Run time
build binary \| windows	2m 57s
build binary \| linux	1m 42s
build binary \| macos x86_64	2m 17s
build binary \| macos aarch64	1m 30s
cargo clippy \| ubuntu	55s
cargo fmt	39s
cargo test \| ubuntu	4m 20s
cargo test \| windows	8m 19s
cargo clippy \| windows	2m 39s
cargo test \| macos	4m 26s
check windows trampoline	4m 7s
check system \| python on macos aarch64	14s
check cache \| macos aarch64	9s
check system \| homebrew python on macos aarch64	22s
check cache \| ubuntu	44s
check system \| amazonlinux	26s
check system \| alpine	26s
ecosystem test \| prefecthq/prefect	1m 53s
check system \| python on debian	44s
ecosystem test \| pallets/flask	10s
check system \| python on fedora	28s
check system \| python on opensuse	33s
check system \| python via pyenv	1m 50s
check system \| pyston	4m 4s
check system \| pypy on ubuntu	31s
check system \| python on rocky linux	38s
check system \| python on ubuntu	16s
check system \| python on macos x86_64	24s
check system \| python3.10 on windows x86	56s
check system \| python3.13 on windows	1m 40s
check system \| python3.12 via chocolatey	1m 33s
check system \| conda3.8 on linux	39s
check system \| python3.10 on windows	51s
check system \| conda3.11 on linux	38s
check system \| conda3.8 on windows	2m 15s
check system \| conda3.11 on windows	1m 52s
check system \| conda3.8 on macos	41s
check system \| conda3.11 on macos	42s
	59m 30s

Provide feedback