feat: support v1.0 and v2.0 notebooks

Turns out, v1 and v2 only differ in how Jupyter interprets them,
not in the schema itself. We can use the same decoder we use
for v3.

decode/decode_test.go

@@ -113,6 +113,36 @@ func TestDecodeBytes(t *testing.T) {
 				}`,
 				nCells: 3,
 			},
+			{
+				name: "v2.0",
+				json: `{
+					"nbformat": 2, "nbformat_minor": 0, "metadata": {}, "worksheets": [
+						{"cells": [
+							{"cell_type": "markdown", "metadata": {}, "source": []},
+							{"cell_type": "markdown", "metadata": {}, "source": []}
+						]},
+						{"cells": [
+							{"cell_type": "markdown", "metadata": {}, "source": []}
+						]}
+					]
+				}`,
+				nCells: 3,
+			},
+			{
+				name: "v1.0",
+				json: `{
+					"nbformat": 1, "nbformat_minor": 0, "metadata": {}, "worksheets": [
+						{"cells": [
+							{"cell_type": "markdown", "metadata": {}, "source": []},
+							{"cell_type": "markdown", "metadata": {}, "source": []}
+						]},
+						{"cells": [
+							{"cell_type": "markdown", "metadata": {}, "source": []}
+						]}
+					]
+				}`,
+				nCells: 3,
+			},
 		} {
 			t.Run(tt.name, func(t *testing.T) {
 				nb, err := decode.Bytes([]byte(tt.json))

schema/v3/schema.go

@@ -20,9 +20,11 @@ import (
 
 func init() {
 	decode.RegisterDecoder(schema.Version{Major: 3, Minor: 0}, new(decoder))
+	decode.RegisterDecoder(schema.Version{Major: 2, Minor: 0}, new(decoder))
+	decode.RegisterDecoder(schema.Version{Major: 1, Minor: 0}, new(decoder))
 }
 
-// decoder decodes cell contents and metadata for nbformat v3.0.
+// decoder decodes cell contents and metadata for nbformat v3.0, v2.0, and v1.0.
 type decoder struct{}
 
 var _ decode.Decoder = (*decoder)(nil)