fix out of range in semi_sync_master_timeout

[luisyonaldo]

Solves issue: #1205

Description

This PR sets semi_sync_master_timeout as bigint to avoid out of range (overflow) errors.

[shlomi-noach]

thank you, a couple changes requested

[shlomi-noach]

(you can see all CI jobs failed -- that's because of the sqlite3 missing hint)

[shlomi-noach]

we need to further change the matching go variable to int64 or uint64, correct? Let's do this in this PR, too

[shlomi-noach]

I think last requested changes

[luisyonaldo]

I think last requested changes

I didn't do uint64 because is not supported by sqlutils library, and all other cases that use bigint unsigned have int64 as counterpart in go

[shlomi-noach]

there is no GetUint64

Whoops. But I think we must use it, right? Because the value reported is 2^64-1, correct?

[shlomi-noach]

I know why sqlutils dosn't have GetUint64 -- that's because golang's ParseInt doen't support parsing a uint64.

[shlomi-noach]

I wonder if we should literally first compare the string with 18446744073709551615

[luisyonaldo]

I know why sqlutils dosn't have GetUint64 -- that's because golang's ParseInt doen't support parsing a uint64.

it does, ParseUint

[shlomi-noach]

it does, ParseUint

!!!

The name confused me, I assumed it returns a uint, not uint64. So inconsistent

[shlomi-noach]

See #1209

[luisyonaldo]

See #1209

Looks good, would be good if we can make #1104 work, so we can backport all this changes to their own repo.

Will wait for #1209 to be merged and then apply changes here

[shlomi-noach]

GetUint64 is now available

[shlomi-noach]

Looks good, would be good if we can make #1104 work

Yeah, I'm 💯 👍 on that. sqlutils has diverged a lot from the original codebase; iirc there's a bit of mess because I also made local changes to sqlutils.go in https://github.com/github/gh-ost