P2: Standardize error handling patterns

[BadgerOps]

Inconsistent Error Handling Pattern

Priority: P2 - Medium
Estimate: 2 hours
Review Reference: REVIEW.md (P2-2)

Issue

Store interface methods use inconsistent error handling patterns:

// Three different patterns:
GetPool(ctx, id) (Pool, bool, error)      // found flag + error
DeletePool(ctx, id) (bool, error)         // success flag + error
CreatePool(ctx, in) (Pool, error)         // just result + error

Impact: Low - Code consistency, but confusing for new contributors

Solution

Standardize to idiomatic Go:

• For Get operations: use (T, error) with ErrNotFound sentinel error
• For Delete: use error only (return ErrNotFound if applicable)
• Define errors in domain package: ErrNotFound, ErrHasChildren, ErrInUse

Example:

// internal/domain/errors.go
var (
    ErrNotFound    = errors.New("not found")
    ErrHasChildren = errors.New("has child resources")
    ErrInUse       = errors.New("resource in use")
)

// Updated interface
func (m *MemoryStore) GetPool(ctx context.Context, id int64) (domain.Pool, error) {
    m.mu.RLock()
    defer m.mu.RUnlock()
    p, ok := m.pools[id]
    if !ok {
        return domain.Pool{}, domain.ErrNotFound
    }
    return p, nil
}

Acceptance Criteria

• domain/errors.go created with sentinel errors
• Store interface updated to use consistent patterns
• MemoryStore updated
• SQLite store updated
• Handlers updated to check errors with errors.Is()
• All tests updated
• Tests verify sentinel error checking

Files Affected

• internal/domain/errors.go (new)
• internal/storage/store.go
• internal/storage/sqlite/sqlite.go
• internal/http/server.go (handler updates)
• All test files

Validation

make test
make lint