Check to see if the project exists before creating keys (#99)

Without this change if a user tries to create a Sentry key for a project
that doesn't exist then they get a not very useful error message from
`terraform apply`. For example, a Terraform resource that looks
something like this:

```
resource "sentry_key" "test" {
  organization = "my-org"
  project      = "project-that-does-not-exist"
  name         = "test-sentry-fix"
}
```

will produce this when `terraform apply` runs:

```
Error: EOF
```

If you set `TF_LOG=trace` then you get more info but still nothing
particularly helpful:

```
apply errored, but we're indicating that via the Error pointer rather than returning it: EOF
```

This is because `resourceSentryKey` assumes that the project exists and
that a 404 from the Sentry API must indicate that the key doesn't exist
and should be created. However, if the project doesn't exist then we
also get a 404 from the Sentry API and any attempt to create a key in
that project that doesn't exist will fail.

With this change the above error will be printed:

```
Error: project not found "project-that-does-not-exist": %!w(<nil>)
```

In this specific case the error variable is `nil` so it's a bit weird
but I think it's useful to display it for situations where it might not
be `nil`.

sentry/resource_sentry_key.go

@@ -1,6 +1,8 @@
 package sentry
 
 import (
+	"fmt"
+
 	"github.com/hashicorp/terraform-plugin-sdk/v2/helper/schema"
 	"github.com/jianyuan/go-sentry/sentry"
 )
@@ -78,6 +80,12 @@ func resourceSentryKeyCreate(d *schema.ResourceData, meta interface{}) error {
 
 	org := d.Get("organization").(string)
 	project := d.Get("project").(string)
+
+	_, resp, err := client.Projects.Get(org, project)
+	if found, err := checkClientGet(resp, err, d); !found {
+		return fmt.Errorf("project not found \"%v\": %w", project, err)
+	}
+
 	params := &sentry.CreateProjectKeyParams{
 		Name: d.Get("name").(string),
 		RateLimit: &sentry.ProjectKeyRateLimit{