pref: remove await/async from internal calls

[harttle]

Remove await/async keywords for internal function calls, so we can remove redundent async/sync implementations (see: #164) and reduce bundle size. This refactor improves performance (both ops/sec and mem used) considerably:

Before refactor:
--- output ---
literal x 16,091 ops/sec ±6.03% (76 runs sampled)
truncate x 21,693 ops/sec ±4.97% (72 runs sampled)
date x 20,426 ops/sec ±4.66% (77 runs sampled)
escape x 25,261 ops/sec ±5.09% (76 runs sampled)
default x 22,055 ops/sec ±4.74% (76 runs sampled)
--- tag ---
if x 20,408 ops/sec ±4.98% (74 runs sampled)
unless x 19,867 ops/sec ±4.87% (77 runs sampled)
for x 8,185 ops/sec ±5.21% (73 runs sampled)
switch x 9,488 ops/sec ±6.23% (73 runs sampled)
assign x 25,552 ops/sec ±5.10% (79 runs sampled)
capture x 27,401 ops/sec ±6.08% (75 runs sampled)
increment x 50,359 ops/sec ±4.99% (78 runs sampled)
decrement x 50,226 ops/sec ±5.80% (74 runs sampled)
tablerow x 9,732 ops/sec ±5.75% (74 runs sampled)
--- demo ---
demo x 3,892 ops/sec ±5.62% (77 runs sampled)
--- layout ---
cache=false x 3,658 ops/sec ±5.70% (69 runs sampled)
cache=true x 8,815 ops/sec ±4.77% (77 runs sampled)
--- memory ---
retained memory for a 3KB template: 16.781KB (250 samples)

After refactor:
--- output ---
literal x 55,118 ops/sec ±2.82% (79 runs sampled)
truncate x 54,365 ops/sec ±2.79% (79 runs sampled)
date x 51,431 ops/sec ±2.59% (80 runs sampled)
escape x 61,429 ops/sec ±3.00% (81 runs sampled)
default x 57,999 ops/sec ±2.77% (76 runs sampled)
--- tag ---
if x 55,281 ops/sec ±2.59% (82 runs sampled)
unless x 47,726 ops/sec ±3.64% (78 runs sampled)
for x 27,228 ops/sec ±2.29% (83 runs sampled)
switch x 28,609 ops/sec ±3.63% (81 runs sampled)
assign x 67,104 ops/sec ±3.71% (78 runs sampled)
capture x 64,171 ops/sec ±2.98% (79 runs sampled)
increment x 95,346 ops/sec ±3.68% (79 runs sampled)
decrement x 92,262 ops/sec ±3.93% (74 runs sampled)
tablerow x 33,876 ops/sec ±3.14% (79 runs sampled)
--- demo ---
demo x 11,217 ops/sec ±2.28% (80 runs sampled)
--- layout ---
cache=false x 6,194 ops/sec ±2.64% (73 runs sampled)
cache=true x 25,483 ops/sec ±2.53% (82 runs sampled)
--- memory ---
retained memory for a 3KB template: 5.30240625KB (250 samples)

In summary, on MacBook 11,1 (2.4 GHz Core i5, 8G RAM) improves include:

• ops/sec for typical use cases improved by x2.5 (see above)
• memory usage reduced by 2/2 (see above)
• size of liquidjs.min.js reduced by 14% (50k -> 43k)

[harttle]

🎉 This PR is included in version 9.2.0 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀