fix get-db-dump

[tobiasmcnulty]

As a workaround for #47 we can simply evaluate the variable in the remote container instead of transferring it locally first.

Fixes this error:

$ inv staging pod.get-db-dump --db-var="DATABASE_URL"
kubectl --namespace <snip> exec -i deploy/web -- pg_dump -Fc --no-owner --clean --dbname W1020 20:38:47.334596 1586748 gcp.go:119] WARNING: the gcp auth plugin is deprecated in v1.22+, unavailable in v1.26+; use gcloud instead.
To learn more, consult https://cloud.google.com/blog/products/containers-kubernetes/kubectl-auth-changes-in-gke
postgres://<snip> > caktus-website-staging_database.dump
W1020 20:38:47.959972 1586794 gcp.go:119] WARNING: the gcp auth plugin is deprecated in v1.22+, unavailable in v1.26+; use gcloud instead.
To learn more, consult https://cloud.google.com/blog/products/containers-kubernetes/kubectl-auth-changes-in-gke
pg_dump: error: too many command-line arguments (first is "20:38:47.334596")
Try "pg_dump --help" for more information.
command terminated with exit code 1
/bin/bash: use: command not found
/bin/bash: line 1: To: command not found
/bin/bash: line 2: postgres://<snip>: No such file or directory

[copelco]

Does this work? The output has an error:

pg_dump: error: too many command-line arguments (first is "20:38:47.334596")

[tobiasmcnulty]

Does this work? The output has an error:

pg_dump: error: too many command-line arguments (first is "20:38:47.334596")

That's the error without the fix (stderr from kubectl is being passed to pg_dump).