flow: rename buffer-length option to buffer-size

It is a better match. I was worried people would understand the size
as in bytes, but we also say a channel size.

docs/configuration.md

@@ -75,7 +75,7 @@ configuration keys:
 - `listen` to specify the IP and UDP port to listen for new flows
 - `workers` to specify the number of workers to spawn to handle
   incoming flows
-- `bufferlength` to specify the number of flows to buffer when pushing
+- `buffer-size` to specify the number of flows to buffer when pushing
   them to the core component
 
 For example:

docs/design.md

@@ -216,6 +216,11 @@ In the future, we may:
 - Add more information to the landing page, including some basic statistics.
 - Automatically build dashboards for Grafana.[^grafana]
 - Builds dashboards with [D3.js][].[^d3js]
+- Buffer message to disks instead of blocking (when sending to Kafka)
+  or dropping (when querying the SNMP poller). At least, we should
+  have a solution to clearly signal when something is not scaling as
+  intended, otherwise, we just drop buffers at the beginning of the
+  pipeline.
 - Collect routes by integrating GoBGP. This is low priority if we
   consider information from Maxmind good enough for our use.