refactor writeto db to smaller function

[yokofly]

I intend to split as 2 prs for #35
the first part is to refactor the writetodb function, I have run some tests, probably the CI will run a more compatible suite.

I conduct a more direct test to verify the temp table: insert a big table, stop with ctrl+c during the insert, and collect the log
this is the whole log difference(between release 1.2.20 and commit ac8802a), from the log seems all good
https://www.diffchecker.com/f4YCxD9W/

[flarco]

Thanks, started to test locally but dealing with another issue.
I'm travelling this week so I may not get to it till this weekend or next week.

Also, there are a few instances of the use of g.Error that need fixing.

For example, instead of g.Error("Could not perform upsert from temp: %v", err) could you use g.Error(err, "Could not perform upsert from temp")? It's a wrapper that I use for error logging/stack tracing, if the first param is anerror type, it treats it differently.

[yokofly]

hahaha, let's enjoy the trip first!
Will adjust the error wrapper.

[yokofly]

I locally performed some simple tests with clickhouse to clickhouse

[flarco]

Tested, all good 👍 .

[yokofly]

this code is buggy, sorry about that, I find the err will pass to the internal, and got something like

WRN NewError called with nil error:

old code

	tCnt, _ := tgtConn.GetCount(tableTmp.FullName())
	if cnt != tCnt {
		err = g.Error("inserted in temp table but table count (%d) != stream count (%d). Records missing/mismatch. Aborting", tCnt, cnt)
		return
	} else if tCnt == 0 && len(sampleData.Rows) > 0 {
		err = g.Error("Loaded 0 records while sample data has %d records. Exiting.", len(sampleData.Rows))
		return
	}