Add Fail fast option in ProcessAll, so that file write errors can be fatal.

[gfr10598]

When we move to GCS json output, we need to be able to terminate parsing if there is a write error.

This also migrates some APIs to use stiface.Client to allow future injection of fakes.

---

This change is 

[coveralls]

Pull Request Test Coverage Report for Build 5952

• 1 of 3 (33.33%) changed or added relevant lines in 1 file are covered.
• 5 unchanged lines in 2 files lost coverage.
• Overall coverage decreased (-0.07%) to 66.816%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
task/task.go	1	3	33.33%

Files with Coverage Reduction	New Missed Lines	%
task/task.go	1	77.12%
active/active.go	4	87.64%

Totals
Change from base Build 5950:	-0.07%
Covered Lines:	3129
Relevant Lines:	4683

---

💛 - Coveralls

[stephen-soltesz]

- with one blocking comment.

Reviewed 6 of 7 files at r1.
Reviewable status: complete! 1 of 1 approvals obtained (waiting on @gfr10598 and @stephen-soltesz)

---

storage/storage.go, line 61 at r1 (raw file):

			return nil, false, err
		} else if strings.Contains(err.Error(), "unexpected EOF") {
			// We are seeing these metrics, but logs at 113

I don't understand this comment.

---

storage/storage.go, line 365 at r1 (raw file):

		return rdr, 0, err
	}
	attr, err := obj.Attrs(ctx)

I added this comment in the draft PR. #902

This is a bug. attr is not guaranteed to be valid when err != nil.

[gfr10598]

Reviewable status: complete! 1 of 1 approvals obtained (waiting on @gfr10598 and @stephen-soltesz)

---

storage/storage.go, line 61 at r1 (raw file):

Previously, stephen-soltesz (Stephen Soltesz) wrote…

I don't understand this comment.

I think I was mistaken. I've improved the comments near line 113, and change next to nextHeader here. I think I may have been confused because the metric labels weren't distinct before.

[gfr10598]

Reviewable status: complete! 1 of 1 approvals obtained (waiting on @stephen-soltesz)

---

storage/storage.go, line 365 at r1 (raw file):

Previously, stephen-soltesz (Stephen Soltesz) wrote…

I added this comment in the draft PR. #902

This is a bug. attr is not guaranteed to be valid when err != nil.

Done.