litebastion: "HTTP/2 transport error" with type=recv_rststream_INTERNAL_ERROR

See https://git.glasklar.is/sigsum/core/log-go/-/issues/101 for context.

We are getting this kind of error messages from litebastion v0.4.1:
```
jan 24 13:42:43 bastion-01 litebastion[70481]: time=2025-01-24T13:42:43.484+01:00 level=INFO msg="HTTP/2 transport error" backend=c77fe626a6b4d53738a9a37920095ee205eb48d1717c19092c4c25efe2f2cc50 remote=[...]:31593 type=recv_rststream_INTERNAL_ERROR
jan 24 13:47:43 bastion-01 litebastion[70481]: time=2025-01-24T13:47:43.552+01:00 level=INFO msg="HTTP/2 transport error" backend=c77fe626a6b4d53738a9a37920095ee205eb48d1717c19092c4c25efe2f2cc50 remote=[...]:31593 type=recv_rststream_INTERNAL_ERROR
jan 24 13:50:43 bastion-01 litebastion[70481]: time=2025-01-24T13:50:43.746+01:00 level=INFO msg="HTTP/2 transport error" backend=c77fe626a6b4d53738a9a37920095ee205eb48d1717c19092c4c25efe2f2cc50 remote=[...]:31593 type=recv_rststream_INTERNAL_ERROR
jan 24 13:52:44 bastion-01 litebastion[70481]: time=2025-01-24T13:52:44.430+01:00 level=INFO msg="HTTP/2 transport error" backend=c77fe626a6b4d53738a9a37920095ee205eb48d1717c19092c4c25efe2f2cc50 remote=[...]:31593 type=recv_rststream_INTERNAL_ERROR
jan 24 13:54:44 bastion-01 litebastion[70481]: time=2025-01-24T13:54:44.448+01:00 level=INFO msg="HTTP/2 transport error" backend=c77fe626a6b4d53738a9a37920095ee205eb48d1717c19092c4c25efe2f2cc50 remote=[...]:31593 type=recv_rststream_INTERNAL_ERROR
jan 24 13:58:13 bastion-01 litebastion[70481]: time=2025-01-24T13:58:13.560+01:00 level=INFO msg="HTTP/2 transport error" backend=c77fe626a6b4d53738a9a37920095ee205eb48d1717c19092c4c25efe2f2cc50 remote=[...]:31593 type=recv_rststream_INTERNAL_ERROR
```

Does this mean that the `recv_rststream_INTERNAL_ERROR` error happened within the litebastion process itself, or could it be an error that happened at the remote system and only the error report got relayed to litebastion?

Is there something that could be done in litebastion to get more information about this kind of error?

Should this be seen as a problem with litebastion itself, or its dependencies, or does it indicate that the remote (the witness) is doing something wrong?

We can reproduce the problem; although we cannot predict exactly when it will happen, it seems to always appear at some point within 24 hours after restarting the sigsum log.

Start by tracing litebastion's HTTP/2 transport error reporting and reproduce the issue described in the logs, using the linked sigsum/core issue for context. Determine whether the reset is local or relayed from the witness, then define what additional diagnostic information and tests would demonstrate the completed fix.

Written by the indexing model from the issue text.