This is related to https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/issues/40230. The initial MR that closed that issue, https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/merge_requests/391, was not semantically correct, because `DataChannelTimeout` starts after the client has already received the answer. After https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/merge_requests/498#note_3156256 the code became not only semantically incorrect, but also functionally incorrect because now if this timeout is hit by the proxy, the client is guaranteed to be gone already. This commit fixes it, by lowering the timeout. This addresses a suggestion in https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/issues/40447. This also closes https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/issues/40381 and supersedes https://gitlab.torproject.org/tpo/anti-censorship/pluggable-transports/snowflake/-/merge_requests/415.