Remove empty lines from changelog before parsing

[sandrofigo]

Remove empty lines from the already read changelog lines before parsing it further in ReadReleaseNotes, similar to the already existing code in ExtractChangelogSectionNotes.

ReadReleaseNotes now accepts an AbsolutePath instead of a string as input, but it should make no difference to the public interface as AbsolutePath has an implicit conversion operator for string.

Closes #1094

I confirm that the pull-request:

• Follows the contribution guidelines
• Is based on my own work
• Is in compliance with my employer

[matkoch]

Changing the signature is not a good idea :) I reverted that part, otherwise thanks for the PR!

[matkoch]

Unfortunately this breaks in the following scenario:

    AbsolutePath ChangelogFile => WorkingDirectory / "CHANGELOG.md";
    ChangeLog Changelog => ReadChangelog(ChangelogFile);

    Target UpdateChangelog => _ => _
        .OnlyWhenStatic(() => Changelog.Unreleased == null)
        .Executes(() =>
        {
            var products = new[]
            {
                HasReSharperImplementation ? "ReSharper" : null,
                HasRiderImplementation ? "Rider" : null
            }.WhereNotNull();

            var changelogLines = ChangelogFile.ReadAllLines().ToList();
            changelogLines.InsertRange(
                Changelog.ReleaseNotes[^1].StartIndex,
                collection: new[]
                {
                    $"## vNext",
                    $"- Added support for {products.Join(" and ")} {ReSharperShortVersion}",
                    string.Empty
                });
            ChangelogFile.WriteAllLines(changelogLines);
        });

I guess the start indexes don't align anymore.

[sandrofigo]

Seems like I have overlooked that, since in my mind the 'StartIndex' property was private the whole time 😓
The property StartIndex refers to the index in the changelog file (without empty lines). I see that the start index is now off due to this PR.
I think we would need to rewrite the changelog parsing a bit more to have the line indices preserved.
This would also affect the unit tests (mostly the ones using Verify) proposed in #1122, so they would need to be adapted as well after this.

Is the StartIndex and EndIndex something vital that needs to be public or could we potentially make it private?