git/blob: GetContentBase64 with fewer allocations and no goroutine (#8297)

See #8222 for context.i

`GetBlobContentBase64` was using a pipe and a goroutine to read the blob content as base64. This can be replace by a pre-allocated buffer and a direct copy.

Note that although similar to `GetBlobContent`, it does not truncate the content if the blob size is over the limit (but returns an error). I think that `GetBlobContent` should adopt the same behavior at some point (error instead of truncating).

### Tests

- I added test coverage for Go changes...
  - [x] in their respective `*_test.go` for unit tests.
- [x] I did not document these changes and I do not expect someone else to do it.

### Release notes

- [x] I do not want this change to show in the release notes.

Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/8297
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>
Co-authored-by: oliverpool <git@olivier.pfad.fr>
Co-committed-by: oliverpool <git@olivier.pfad.fr>

modules/git/blob_test.go

@@ -63,6 +63,24 @@ func TestBlob(t *testing.T) {
 		require.Equal(t, "file2\n", r)
 	})
 
+	t.Run("GetContentBase64", func(t *testing.T) {
+		r, err := testBlob.GetContentBase64(100)
+		require.NoError(t, err)
+		require.Equal(t, "ZmlsZTIK", r)
+
+		r, err = testBlob.GetContentBase64(-1)
+		require.ErrorAs(t, err, &BlobTooLargeError{})
+		require.Empty(t, r)
+
+		r, err = testBlob.GetContentBase64(4)
+		require.ErrorAs(t, err, &BlobTooLargeError{})
+		require.Empty(t, r)
+
+		r, err = testBlob.GetContentBase64(6)
+		require.NoError(t, err)
+		require.Equal(t, "ZmlsZTIK", r)
+	})
+
 	t.Run("NewTruncatedReader", func(t *testing.T) {
 		// read fewer than available
 		rc, size, err := testBlob.NewTruncatedReader(100)