Describe the bug
Tailing a deployment with multiple pods results in stream failed

To Reproduce
Steps to reproduce the behavior:

1. Select deployment with multiple pods (6 in my case)
2. Use the l character to start log tailing.
3. Observe logs stream for a time then bail out with stream failed

Expected behavior
Logs continue streaming until I give up.

Screenshots
Here are the last few lines in the output. Note also that newlines are not being inserted in front of some lines of output:

mainstream-dequeue-5cd4cffb96-vht7k:2020-07-03 02:08:21 DEBUG MainStreamDequeue: processed perf: a466c83954a1559dd343fd4a6e73cf08 1593742008258
mainstream-dequeue-5cd4cffb96-2pknp:2020-07-03 02:08:21 DEBUG MainStreamDequeue: processed perf: e692822f63da8f92760af710cd75843c 1593741690970
mainstream-dequeue-5cd4cffb96-4sss2:2020-07-03 02:08:21 DEBUG MainStreamDequeue: processed perf: 9155b83c56bc2621e3f83c22de1dcc43 1593741956122mainstream-dequeue-5cd4cffb96-f4brn:2020-07-03 02:08:21 DEBUG MainStreamDequeue: processe
mainstream-dequeue-5cd4cffb96-2v5p6:2020-07-03 02:08:21 DEBUG MainStreamDequeue: processed perf: 7659c9f7117904fca77f15681936a232 1593741994508mainstream-dequeue-5cd4cffb96-4xntw:2020-07-03 02:08:21 DEBUG MainStreamDequeue: processe
mainstream-dequeue-5cd4cffb96-2pknp:stream failedmainstream-dequeue-5cd4cffb96-4xntw:stream failed

Note that the messages are truncated and several may occur on one line.

Versions (please complete the following information):

• OS: OSX 10.13.6
• K9s 0.21.0 [59018]
• K8s v1.16.8-gke.15

Additional context
Note that in the previous version 0.20.5 it would hang and require a kill -9 so this is an improvement since now an escape will return.
In 0.20.5 though the newlines were properly included with each log line.
Perhaps this problem is creating too long a buffer for lines?