chore: document migration from TravisCI to GitHub Actions

Relate to #1154

[skip ci]

Author: php-coder

docs/decisions-log.md

@@ -2,6 +2,7 @@
 
 | Date/Type | Description |
 | --------- | ----------- |
+| 20.07.2023 Decision | [php-coder/mystamps#1154](https://github.com/php-coder/mystamps/issues/1154): some time ago (29.11.2022) **Travis CI** [stopped to work](https://github.com/php-coder/mystamps/issues/1154#issuecomment-1330014938) because of the changes in their policy. On 04.12.2022 [I contacted with support and got 25k credits](https://github.com/php-coder/mystamps/issues/1154#issuecomment-1336413246) and CI has started to work again. This was a signal to try **GitHub Actions** as it's more stable, feature-rich and native for GitHub. Today (20.07.2023) a migration has finished and the project now is fully moved to GitHub Actions |
 | 22.06.2023 Workaround | [php-coder/mystamps#1610](https://github.com/php-coder/mystamps/issues/1610): in the `connect-todos-to-issues.sh` script, we want to search issues that a) contains puzzle id in their body b) has a title like we have in a puzzle. The former condition covers the case when a puzzle id already has a related issue. The latter is for the case when puzzle id got changed but the title remains the same (for example, because of changing an intepretation of the comments) -- this is how we try to find an issue that was connected to an old puzzle. The limitation that I faced is that **`gh`** utility can't use *logical OR* in the search conditions. I added a workaround where we first search in body, after that by title, and at the end we merge the results together. Commit: [c02dc39a](https://github.com/php-coder/mystamps/commit/c02dc39a7357d30e8c8f74c346e5fc4b81122140) |
 | 22.06.2023 Workaround | [php-coder/mystamps#1610](https://github.com/php-coder/mystamps/issues/1610): in the `connect-todos-to-issues.sh` script, we use [**`gh`** cli tool](https://cli.github.com) for searching issues by its title. But it's turned out that it searches issues with titles that contain a phrase while we need an *exact match*. Discussion: [link](https://github.com/orgs/community/discussions/17956) Commit: [c02dc39a](https://github.com/php-coder/mystamps/commit/c02dc39a7357d30e8c8f74c346e5fc4b81122140)<br />UPDATE(28.06.2023): I'm going to add a workaround for that by filtering out the titles within the script. Commit: [1c5c0c2b](https://github.com/php-coder/mystamps/commit/1c5c0c2bd94d083114e05f0d43d4bfb727e84125) Issue: [#1618](https://github.com/php-coder/mystamps/issues/1618) |
 | 16.06.2023 Workaround | [php-coder/mystamps#1610](https://github.com/php-coder/mystamps/issues/1610): in one of the releases **pdd** has changed it's behavior for parsing of the multiline comments: previously, in order to be included in a title, the next line should have started from a space. The current behavior is that pdd reads all the next lines until a comment ends or the next puzzle appears. I had to modify the existing comments in order to make their titles look as expected. I also submitted an issue to the upstream in order to improve documentation regarding these rules. Commits: [c2adb2aa](https://github.com/php-coder/mystamps/commit/c2adb2aad0ea58a1aee85cbe4aa5838dfb70f510) Issue: [cqfn/pdd#224](https://github.com/cqfn/pdd/issues/224) |