Currently, the timeout is hardcoded with the value of 250ms. However,
when developing, we may want to make the timeout longer to make sure we
do not miss any messages due to other reasons.

This commit adds timeout overriding using program arguments. This makes
tweaking the timeout at runtime possible.

Tested:

1. Send a request to a non-existent endpoint. It should timeout in 250ms
   by default.

       root@mtmitchell:~# time busctl call xyz.openbmc_project.MCTP /xyz/openbmc_project/mctp \
       > au.com.CodeConstruct.MCTP SetupEndpoint say mctppcie0 2 0xCA 0xFE
       Call failed: MCTP Endpoint did not respond

       real    0m0.344s
       user    0m0.016s
       sys     0m0.000s

2. Change mctpd.service to use longer timeout, 10 seconds for example.

       ExecStart=/usr/sbin/mctpd -t 10000000

3. Resend the request. It should timeout in 10 seconds.

       root@mtmitchell:~# time busctl call xyz.openbmc_project.MCTP /xyz/openbmc_project/mctp \
       > au.com.CodeConstruct.MCTP SetupEndpoint say mctppcie0 2 0xCA 0xFE
       Call failed: MCTP Endpoint did not respond

       real    0m10.083s
       user    0m0.015s
       sys     0m0.001s

Signed-off-by: Khang Nguyen <[email protected]>

Currently, mctpd defaults to run as a bus owner, so ctx->bus_owner is
always true. This commit adds a role argument so that we can change the
role without editing the source code.

Tested: Send a Get EID command to mctpd. See the response:

- Running as Bus Owner (no flag):

      00 00 02 00 ee 12 00

- Running as Endpoint (-r endpoint):

      00 00 02 00 ee 02 00

The second to last byte signifies the endpoint type.
(0xee is the EID assigned to mctpd)

Signed-off-by: Khang Nguyen <[email protected]>