Currently we check in many different places whether debug is enabled
prior to raising a debug event.
This changes that to always raise debug, but let the debug.raise()
method only store it on the pattern object if/when debug is
enabled.
## 2.18.0 (2021-09-09)
### core
#### Fixed
- Handle path.offset() of very short curves with control points on the start or end point Closes [#1257](https://github.com/freesewing/freesewing/issues/1257)
### holmes
#### Fixed
- The `brimWidth` option is not a percent option, allowing the pattern to scale properly
### huey
#### Fixed
- Replace cut-on-fold indicator on pocket with a regular grainline indicator Closes [#1265](https://github.com/freesewing/freesewing/issues/1265)
### i18n
#### Added
- Added translations for Yuri
#### Fixed
- Added optional chaining so missing options always lead to clear error message
### simon
#### Fixed
- Avoid paperless depending on a complete pattern
### theo
#### Fixed
- Avoid paperless depending on a complete pattern
### wahid
#### Fixed
- Close Seam Allowance path of front lining Closes [#1267](https://github.com/freesewing/freesewing/issues/1267)
- Support a zero value for the `backScyeDart` option
### yuri
- Handle path.offset() of very short curves with control points on the start or end point Closes [#1257](https://github.com/freesewing/freesewing/issues/1257)
- The `brimWidth` option is not a percent option, allowing the pattern to scale properly
- Replace cut-on-fold indicator on pocket with a regular grainline indicator Closes [#1265](https://github.com/freesewing/freesewing/issues/1265)
- Added translations for Yuri
- Added optional chaining so missing options always lead to clear error message
- Avoid paperless depending on a complete pattern
- Avoid paperless depending on a complete pattern
- Close Seam Allowance path of front lining Closes [#1267](https://github.com/freesewing/freesewing/issues/1267)
- Support a zero value for the `backScyeDart` option
This path.offset method used to travel 2mm along the path to find
out the direction at the start/end of paths.
That caused errors when the curve being offsetted was less then 2mm.
So now, it checks and if it's shorter will use 10% of the path length instead.
This closes#1257
This bug would be triggered by an edge-case where we have a path
that is made up of multiple curves/lines.
If the distance to shift lies just beyond the end of one segment,
yet the remaining distance is smaller than the step (by default
that means smaller than 1/25mm) no point would be returned.
This closes#1140
