There are two things here.

First of all, the fact that parser erases the text comment "French, I thought he would play Carokann" is a bug. This comment has to be rendered in some way. This is a bug.

But on the other hand, putting a text comment just between the end of a sub-variation (i.e. (1...c6)) and the following move (i.e. 2.d4) is not supported by the parser, and that is the expected behavior. The way the parser works, a text comment is always linked:

1. to the previous move in the same variation,
2. or to the beginning of the variation if it appears before the first move of the variation.

For instance, with

1.e4 e6 ({Comment A} 1...c6 {Comment B}) {Comment C} 2.d4

... Comment C is linked to move 1...e6 in the main variation, Comment B is linked to move 1...c6 in the sub-variation, and Comment A is linked to the beginning of the sub-variation. And during the restitution, a text comment always rendered just after the move it relates to (or at the very beginning of the sub-variation if it is linked to a sub-variation).

Hence, in your example, the expected output is:

1.e4 e6 French, I thought he would play Carokann , but fortunately I like French too (1…c6) 2.d4 d5 and white wins

This is what you will obtain with version 7.5.1 (just released).

But considering the meaning of the text comments in this particular example, I would suggest instead to write them as:

1.e4 e6 {French} ({I thought he would play Carokann} 1...c6 {but fortunately I like French too}) 2.d4 d5 {and white wins} *

Thanx for fixing the bug. I can live with the way you implemented it. Some clubmembers are generating pgn via all kind of of (online) programs. Before publishing we have to check the way it looks on the website of the club. No problem and thanx again!