Background services livenessProbe is too aggressive

The background services running for sourcegraph should not require that high request/limits, we are running repo-updater with 50m cpu and 128Mi memory for instance, this makes the container fail to start before the livenessProbe(which will restart it after 3 seconds as current configured) and put the pod into a crash loop.

Setting a intialDelaySeconds to 5s, or startupProbe, or simply making the livenessProbe of the non-user hot path services less aggressive would allow these services to run with less resources.

检查后台服务的 Helm 模板，尤其是 repo-updater，并找出它们的 livenessProbe 和资源设置。将 probe 的计时与所述的 50m CPU 和 128Mi 内存分配进行比较；完成的标准是，渲染后的配置能够让低资源服务启动，而不会进入崩溃循环。