Semantic versioning approach for Wrangler

[petebacondarwin]

In May we decided to modify our use of semantic versioning when publishing Wrangler to npm. The modified version deviates from the semver specification by allowing all non-breaking changes (i.e. bug fixes and small features) to appear in patch releases.

When we implemented this, we agreed that we would see review the approach after a short period. This RFC is a start of a discussion, between the team and the wider community as to whether we keep this approach or revert to the standard common approach.

Advantages of modified approach

• Users of Wrangler are more likely to upgrade to the latest version because patch versions signal a safer upgrade path.
  This frees the Wrangler team to move more quickly when adding new features without the additional overhead of grouping and planning rollout of features.

Advantages of semver

• Wrangler users can rely upon a common understanding of semantic versions to reason about the Wrangler package versions.
  This may mean that developers are more cautious about upgrading to new minor versions but then they can be confident that a patch version release only contains bug fixes.
• Wrangler developers can rely upon a common understanding of semantic versions when deciding how to tag the changesets for the PRs.
  This avoids a layer of documentation and process to prevent unnecessary minor version bumps to the Wrangler package for non-breaking features.

Please add additional thoughts (advantages, disadvantages, etc) below.